Tài liệu Report on The Management of Non-Muscle-Invasive Bladder Cancer ppt

The panel developed practice policy recommendations for three types of patients: 1 a patientwho presents with an abnormal growth on the urothelium, but who has not yet been diagnosed wit

Bladder Cancer Guidelines Panel

Randall G Rowland, MD Abraham T.K Cockett, MD John A Fracchia, MD

Members:

Consultants:

Hanan S Bell, PhD Patrick M Florer Curtis Colby

Bladder Cancer Clinical Guidelines Panel Members and Consultants

The Bladder Cancer Clinical Guidelines Panel consists of board-certified urologists who are experts in the treatment of

blad-der cancer This Report on the Management of Non-Muscle-Invasive Bladblad-der Cancer (stages Ta, T1 and Tis) was extensively

reviewed by over 50 physicians throughout the country in February 1999 The Panel finalized its recommendations for the

American Urological Association (AUA) Practice Parameters, Guidelines and Standards Committee, chaired by Joseph W.

Segura, MD, in July 1999 The AUA Board of Directors approved these practice guidelines in August 1999.

The Summary Report also underwent independent scrutiny by the Editorial Board of the Journal of Urology, was accepted for publication in August 1999, and appeared in its November 1999 issue A Doctor’s Guide for Patients and Evidence Working Papers have also been developed; both are available from the AUA.

The AUA expresses its gratitude for the dedication and leadership demonstrated by the members of the Bladder Cancer

Clinical Guidelines Panel in producing this guideline

Residents (Data Extraction) Jack Baniel

Elie Benaim Clay Gould Blake Hamilton Jeff Holzbeierlien Fred Leach John Mansfield Mitchell S Steiner Brad Stoneking Joseph Trapasso Margaret Wolf

Abraham T.K Cockett, M.D.

(Physician Consultant) Department of Urology University of Rochester Rochester, New York John A Fracchia, M.D.

(Physician Consultant) Chief

Section of Urology Department of Surgery Lenox Hill Hospital New York, New York

Archived Document— For Reference Only

More than 50,000 new bladder cancer cases are diagnosed each year in the United States, andthe incidence rate (number of new cases per 100,000 persons per year) has been slowly rising, con-current with an aging population (Landis, Murray, Bolden, et al., 1999; Parker, Tong, Bolden, etal., 1996, 1997; Wingo, Tong, Bolden, et al., 1995).

Bladder cancer is largely a disease afflicting the late middle age and old age populations

Although the disease does occur in young persons—even in children—more than 70 percent of

new cases are diagnosed in persons aged 65 and older (Lynch and Cohen, 1995; Yancik and Ries,1994) As the baby boom generation ages over the next two decades, the incidence of bladder can-cer will likely accelerate

At any age, most bladder cancers, when initially diagnosed, have not invaded the detrusor cle (Fischer, Waechter, Kraus, et al., 1998; Fleshner, Herr, Stewart, et al 1996) These noninvasive

mus-cancers are the subject of this Report on the Management of Non-Muscle-Invasive Bladder Cancer

(Stages Ta, T1 and Tis) The report was produced by the American Urological Association's

Bladder Cancer Clinical Guidelines Panel

The AUA charged the panel with the task of analyzing published outcomes data to assess tential benefits and possible adverse effects of treatment interventions and to produce practice poli-

po-cy recommendations accordingly The three types of outcomes the panel determined to be mostimportant for analysis are: 1) probability of tumor recurrence; 2) risk for tumor progression; and3) complications of treatment

The panel developed practice policy recommendations for three types of patients: 1) a patientwho presents with an abnormal growth on the urothelium, but who has not yet been diagnosed

with bladder cancer; 2) a patient with established bladder cancer of any grade, stage Ta or T1, with

or without carcinoma in situ (CIS), who has not had prior intravesical therapy; and 3) a patient

with CIS or high-grade T1 cancer who has had at least one course of intravesical therapy

The panel avoided use of the term "superficial" in this report to categorize the three cle-invasive stages of bladder cancer: Ta, T1 and Tis The panel agrees with the International

non-mus-Society of Urological Pathology's recommendation that such use of the term should be discouraged(Epstein, Amin, Reuter, et al., 1998) Ta, T1 and Tis tumors have often been grouped together as

"superficial" cancers because they are all superficial to the detrusor muscle, but in most other spects they behave differently from one another and to group them in a single category is mislead-ing (See the discussion on page 16.)

re-A summary of this report has been published in the Journal of Urology (November 1999) re-A

Doctor's Guide for Patients and Evidence Working Papers are available for purchase through the

AUA

Introduction

Archived Document— For Reference Only

Archived Document— For Reference Only

Introduction i

Executive Summary 1

Methodology 1

Background 1

Treatment alternatives 2

Treatment recommendations 3

Chapter 1: Methodology 9

Literature search, article selection and data extraction 10

Evidence combination 10

Limitations 11

Chapter 2: Non-muscle-invasive bladder cancer and its management 13

Etiology 13

Major types of bladder cancer 13

Histology 14

Diagnosis 14

Staging 15

Grade classification 17

Prognostic indicators 17

Treatment alternatives 18

Follow-up 20

Chapter 3: Outcomes analysis for treatments of non-muscle-invasive bladder cancer 21

The outcome tables 21

Variability of outcomes data 25

Outcomes summary:recurrence and progression 25

Outcomes summary: treatment complications 26

Chapter 4: Recommendations for management of non-muscle-invasive bladder cancer 28

Treatment policies: levels of flexibility 28

Index patients 28

Treatment recommendations 29

Areas for future research 31

References 33

Appendix A – Data Presentation 38

Appendix B – Data extraction form 54

Appendix C – Data analysis 56

Index 58

Contents

Archived Document— For Reference Only

Archived Document— For Reference Only

To develop recommendations for treatment of

non-muscle-invasive bladder cancer, the AUA

Bladder Cancer Clinical Guidelines Panel

re-viewed the literature on bladder cancer from

January 1964 to January 1998 and extracted and

meta-analyzed all relevant data to estimate as

ac-curately as possible both desirable and

undesir-able outcomes of alternative treatment modalities

The panel followed an explicit approach to the

de-velopment of practice policy recommendations

(Eddy, 1992) This approach emphasizes the use

of scientific evidence in estimating outcomes If

the evidence has limitations, the limitations are

clearly stated When panel opinion is necessary,

the explicit approach calls for an explanation of

why it is necessary and/or for discussion of the

factors considered For a full description of the

methodology, see Chapter 1

More than 90 percent of all bladder cancers

in the United States and Europe, both

muscle-invasive and nonmuscle-invasive, are transitional cell

car-cinomas originating in the urothelium that forms

the bladder lining (Fleshner, Herr, Stewart, et al.,

1996) Transitional cell carcinomas may appear in

a variety of configurations—including exophytic

papillary tumors (the most common

configura-tion); flat patches of carcinoma in situ (CIS);

nodular tumors; sessile growths; and mixed

growths such as high-grade papillary tumors

to-gether with flat CIS

Hematuria is the usual first sign of bladder

cancer, present at least microscopically in almost

all patients with cystoscopically detectable tumors

(Messing and Valencourt, 1990) In some cases,

bladder irritability accompanied by urgency,

fre-quency and dysuria will be present in addition to

hematuria This set of symptoms is associatedwith diffuse CIS or muscle-invasive disease

Routine diagnostic methods for bladder cancerinclude: a thorough history, especially regardingexposure to known carcinogens; a physical exami-nation; urine analysis; and a cystoscopic examina-tion of the bladder and urethra Diagnostic cysto-scopies today are usually outpatient proceduresdone with a flexible or rigid cystoscope under lo-cal anesthesia Diagnostic tools available in addi-tion to cystoscopy include cytologic assessmentand several new urinary tests approved by theFDA for detection or monitoring of recurrences ofbladder cancer

A transurethral resection of a bladder tumor(TURBT) is usually performed both to excise allvisible tumors and to provide specimens forpathologic evaluation to determine tumor stageand grade (Shelfo, Brady and Soloway, 1997)

Additional loop or cold-cup biopsies may be

tak-en to evaluate other areas of the urothelium, and abimanual palpation before and after resection mayprovide further information on tumor size anddepth of penetration A repeat TURBT may beperformed in cases of incompletely resected Ta tu-mors and T1 tumors

Staging

For staging of bladder cancer, the originalJewett-Strong system (1946), modified byMarshall (1952, 1956), has generally given way tothe TNM (tumor, node, metastasis) system devel-oped jointly by the American Committee onCancer Staging and the International UnionAgainst Cancer (Hermanek and Sobin, 1992;

Fleming, 1997) Depth of tumor penetration isthe crucial element in both systems Table 1 onpage 16 shows the TNM classifications for prima-

ry tumors, adapted from the American JointCommittee on Cancer (AJCC) staging manual(Fleming, 1997)

Executive Summary:

Report on the management of non-muscle-invasive

bladder cancer (stages Ta, T1 and Tis)

Methodology

Background

Archived Document— For Reference Only

Basic characteristics of

stages Ta, T1 and Tis

Stage Ta tumors are confined to the urothelium

(above the basement membrane) and have a

papil-lary configuration described by Johansson and

Cohen (1996) as resembling "seaweed" protruding

into the lumen of the bladder Most Ta tumors are

low grade

Stage T1 tumors have penetrated below the

basement membrane and infiltrated the lamina

propria, but not so far as the detrusor muscle

Most T1 tumors are papillary, but many of those

that have penetrated the deepest into the lamina

propria are nodular (Heney, Nocks, Daly, et al.,

1982)

In a stage by itself, CIS (stage Tis) has been

defined as high-grade (anaplastic) carcinoma,

which like stage Ta is confined to the urothelium,

but with a flat, disordered, nonpapillary

configura-tion and a likelihood of being underdiagnosed

(Epstein, Amin, Reuter, et al., 1998) CIS can be

focal, multifocal or diffuse On cystoscopic

ex-amination, it usually appears as a slightly raised,

reddened patch of velvety mucosa but often is

endoscopically invisible

Grade classification

Numerous classification systems for grading

transitional cell carcinomas of the bladder have

been developed and published over the past few

decades Although no single system has yet

emerged to win universal acceptance, the most

widely used systems all share important

character-istics In particular, they all tend to group bladder

carcinomas similarly into three principal grades

based mainly on degree of anaplasia (Bergkvist,

Ljungqvist and Moberger, 1965; Epstein, Amin,

Reuter, et al., 1998; Koss, 1975) The three

grades—low (grade 1), intermediate (grade 2) and

high (grade 3)—correspond respectively to well

differentiated, moderately differentiated and

poor-ly differentiated tumors Grade has been shown

to be a highly predictive indicator of future tumor

behavior with regard to both recurrence and

pro-gression

In most cases of non-muscle-invasive bladder

cancer, tumors are treated initially with TURBT,

fulguration and/or laser therapy A careful scopic examination of all bladder surfaces, theurethra and the prostate precedes resection (Kochand Smith, 1996) Findings with prognostic sig-nificance are noted during this examination

cysto-Following resection, adjuvant intravesicalchemotherapy or intravesical immunotherapy iscommonly used to prevent recurrences

Resection and fulguration of

bladder tumors

As stated previously, a TURBT has two mainpurposes: 1) complete eradication of all visible tu-mors; and 2) tissue resection for pathologic evalu-ation to determine grade and stage Fulgurationmay be used on small lesions, but tissue stillneeds to be obtained to determine grade and stage

at time of initial presentation When a tumor isremoved, a separate biopsy can be taken at thebase with the resecting loop, after which healthy-appearing muscle fibers should be visible at thebase Necrotic-appearing tissue implies an inva-sive carcinoma The presence of fat implies afull-thickness bladder wall defect

Laser therapy

The Nd:YAG laser has so far proven to be themost versatile wavelength for treating bladdercancer, but other wavelengths also have been used(Koch and Smith, 1996; Smith, 1986) Resultsare comparable to electrocautery resection, withlittle difference in the recurrence rate (Beislandand Seland, 1986) However, tissue samples need

to be obtained beforehand by means of cold-cupbiopsies to determine tumor grade Assessingdepth of tumor penetration to determine stage ismore problematic with laser therapy Appropriatepatients for this therapy have papillary, low-gradetumors and a history of low-grade, low-stage tu-mors (Koch and Smith, 1996)

Intravesical chemotherapy and

immunotherapy

Intravesical chemotherapy or immunotherapy

is most often used as adjuvant treatment to vent tumor recurrence following the TURBT ofprimary non-muscle-invasive bladder tumors, in-cluding possible recurrence because of iatrogenicimplantation of tumor cells Intravesical therapy

pre-is also used to treat known expre-isting tumors in casesTreatment alternatives

Archived Document— For Reference Only

of CIS, which frequently cannot be treated

ade-quately by resection or fulguration because of

dif-fuse involvement The chief intravesical agents

currently available are thiotepa, doxorubicin,

mit-omycin C and bacillus Calmette-Guérin (BCG)

Chapter 3 of this report contains an

evidence-based comparative outcomes analysis of these

agents

Thiotepa, introduced in 1961, is the oldest and

one of the least expensive of the intravesical

drugs It is an alkylating agent that acts by

cross-linking nucleic acid Its low molecular weight of

189 allows partial absorption through the

urotheli-um, with possible systemic toxicity

Doxorubicin is an anthracycline antibiotic able

to bind to DNA and inhibit synthesis It is not

cell cycle specific, but appears to be most

cytotox-ic in the S phase Its molecular weight of 580 is

high, and absorption and systemic toxicity are

ex-tremely rare

Mitomycin C is an antibiotic that works by

inhibiting DNA synthesis Because of its

moder-ately high molecular weight of 329, there are

few problems with transurothelial absorption,

and myelosuppression is rare However,

mito-mycin C is a very expensive agent (see Table 6

on page 25)

BCG is a live attenuated strain of

Mycobacterium bovis and was first used as a

tu-berculosis vaccine Its now widespread use as

in-travesical immunotherapy for management of

noninvasive bladder cancer began in the 1970s It

has since become a first-line treatment for CIS

and has been shown to be effective as prophylaxis

to prevent bladder cancer recurrences following

TURBT (Cookson and Sarosdy, 1992; Coplen,

Marcus, Myers, et al., 1990; DeJager, Guinan,

Lamm, et al., 1991; Herr, Schwalb, Zhang, et al.,

1995; Lamm, Blumenstein, Crawford, et al.,

1995) Its mechanism of action is not fully

under-stood, but clearly involves a strong inflammatory

immunologic host response with release of

inter-leukins and other cytokines (Morales, Eidinger

and Bruce, 1976; Ratliff, Haaff and Catalona,

1986) The most common side effects of BCG

are cystitis and hematuria The most serious is

BCG sepsis BCG therapy is contraindicated in

patients who are immunocompromised, have liver

disease or a history of tuberculosis

The panel generated its recommendationsbased on analysis of comparative outcomes datafrom both randomized controlled trials (RCTs)and clinical series and on expert opinion Therecommendations apply to treatment of patientswith non-muscle-invasive, transitional cell carci-noma of the bladder, including CIS as well asstages Ta and T1 tumors The panel evaluatedcomparative data for the following treatmentmethods in particular:

• TURBT;

• TURBT plus thiotepa;

• TURBT plus doxorubicin;

• TURBT plus mitomycin C;

• TURBT plus BCG

The terms "standard," "guideline" and tion," as used in the panel's recommendations, re-fer to the three levels of flexibility for practicepolicies defined in Chapter 1 (page 9) A standard

"op-is the least flexible of the three, a guideline moreflexible and an option the most flexible Optionscan exist because of insufficient evidence or be-cause patient preferences are divided In the lattercase particularly, the panel considered it important

to take into account likely preferences of ual patients when selecting from among alterna-tive interventions

individ-Index patients

The specific types of patients to whom thepanel's recommendations apply are termed indexpatients In recognition of the differences in deci-sion-making that occur depending upon patientcircumstances, the panel defined three differentindex patients:

Index Patient No 1: A patient who presents

with an abnormal growth on the urothelium, butwho has not yet been diagnosed with bladder can-cer;

Index Patient No 2: A patient with

estab-lished bladder cancer of any grade, stage Ta orT1, with or without CIS, who has not had priorintravesical therapy; and

Index Patient No 3: A patient with CIS or

high-grade T1 cancer who has had at least onecourse of intravesical therapy

(continued on page 6)

Treatment recommendations

Archived Document— For Reference Only

Recommendation for all index patients

Standard:

Physicians should discuss with the patient the treatment options and the benefits and

harms, including side effects, of intravesical treatment, especially those side effects

associated with a particular agent

Recommendation for Index Patient No 1

A patient who presents with an abnormal growth on the urothelium, but who has not

yet been diagnosed with bladder cancer.

Standard:

If the patient does not have an established histologic diagnosis, a biopsy should be

ob-tained for pathologic analysis

Recommendations for Index Patient No 2

A patient with established bladder cancer of any grade, stage Ta or T1, with or

with-out CIS, who has not had prior intravesical therapy.

Standard:

Complete eradication of all visible tumors should be performed if surgically feasible

and if the patient's medical condition permits

Option:

Surgical eradication can be performed by one of several methods, including

electro-cautery resection, fulguration or laser ablation

(continued on next page)

Archived Document— For Reference Only

Recommendations (continued)

Option:

Adjuvant intravesical chemotherapy or immunotherapy is an option for treatment after

endoscopic removal of low-grade Ta bladder cancers

Guideline:

Intravesical instillation of either BCG or mitomycin C is recommended for treatment

of CIS and for treatment after endoscopic removal of T1 tumors and high-grade Ta

tu-mors

Option:

Cystectomy may be considered for initial therapy in some patients with CIS or T1

tu-mors

Recommendations for Index Patient No 3

A patient with CIS or high-grade T1 cancer who has had at least one course of

intrav-esical therapy.

Option:

Cystectomy may be considered as an option for patients with CIS or high-grade T1

cancers that have persisted or recurred after an initial intravesical treatment

Option:

Further intravesical therapy may be considered as an option for patients with CIS or

high-grade T1 cancers that have persisted or recurred after an initial intravesical

treat-ment

Archived Document— For Reference Only

Recommendation for all

index patients

This recommendation is based on the panel's

expert opinion Although all of the adjuvant

treat-ments studied decrease the recurrence probability

of bladder cancer when they are compared to

TURBT alone, the published literature does not

unequivocally support the conclusion that the rate

of progression to muscle-invasive disease is

al-tered Patients contemplating intravesical therapy

should consider information about all expected

outcomes including the possible side effects of

therapy All of the treatments have in common

some side effects, especially those related to

blad-der irritation; other side effects are either unique

or peculiar to a particular drug The incidence of

serious or life-threatening side effects with any of

the treatments is low, as shown in Table 5; but the

panel is aware of reports documenting systemic

in-fection or even death from BCG sepsis in rare

cir-cumstances

In addition, there is little information defining

the optimal dose, number and timing of

instilla-tions or the influence of long-term maintenance

therapy Randomized studies supporting the use

of maintenance BCG therapy have been

conduct-ed, but the panel excluded them because their

re-sults have been reported only in abstract form

Other trials would appear to support a role for

maintenance therapy

Recommendation for

Index Patient No 1

A patient who presents with an abnormal

growth on the urothelium, but who has not yet

been diagnosed with bladder cancer.

Although urine cytology and other novel tumormarkers, as discussed in Chapter 2 (page 17), canprovide suggestive evidence of bladder cancer, thedefinitive diagnosis is established by pathologicexamination of tissue removed by TURBT orbiopsy Transitional cell carcinoma of the bladderoften has a characteristic appearance, but otherconditions can mimic the gross appearance ofbladder cancer It is important, therefore, to prove

by microscopic analysis that gross abnormalitiesare from transitional cell carcinoma Intravesicalimmunotherapy or chemotherapy should not beused in the absence of a histologic diagnosis

Recommendations for Index Patient No 2

A patient with established bladder cancer of any grade, stage Ta or T1, with or without CIS, who has not had prior intravesical therapy.

In a patient with known bladder cancer, scopic ablation of all visible tumors should be per-formed This can be accomplished with electro-cautery resection, fulguration or application oflaser energy Sometimes, the extent of disease orthe location of the lesions may preclude completeeradication Also, co-morbid conditions must beconsidered and may occasionally influence a deci-sion about whether or not to attempt endoscopicremoval of bladder tumors This recommendation

endo-is based on the panel's expert opinion

For bladder cancers that do not invade the trusor muscle, several methods for tumor destruc-tion provide apparently comparable results

de-Adequate tissue should be available for tion of pathologic stage, but endoscopic ablative

determina-Standard: Physicians should discuss with the

patient the treatment options and the benefits

and harms, including side effects, of

intravesi-cal treatment, especially those side effects

asso-ciated with a particular agent

Standard: If the patient does not have an

es-tablished histologic diagnosis, a biopsy should

be obtained for pathologic analysis

Standard: Complete eradication of all visible

tumors should be performed if surgicallyfeasible and if the patient's medical conditionpermits

Option: Surgical eradication can be performed

by one of several methods, including cautery resection, fulguration or laser ablation

electro-Archived Document— For Reference Only

techniques may not permit submission of all

ma-terial for histologic evaluation This

recommen-dation is based on the panel's expert opinion

A recommendation stronger than an option for

this group of patients (Index Patient No 2) cannot

be supported by the evidence in the outcomes

ta-bles (see page 23) Most studies combined

pa-tients who had low-grade stage Ta tumors with

patients who had T1 lesions and higher-grade

can-cers, creating difficulty in extracting information

about this subgroup However, based upon panel

opinion, many patients with low-grade Ta tumors

do not require adjuvant intravesical therapy

There is a low risk of disease progression (overall

less than 10 percent) in this group, and there is

lit-tle evidence that adjuvant therapy affects the

pro-gression rate For rapidly recurring low-grade Ta

tumors, adjuvant therapy may be useful The

out-comes tables show a decreased recurrence

proba-bility for all the intravesical therapies studied,

compared to TURBT alone

Based on evidence from the literature and

pan-el opinion, both BCG and mitomycin C are

supe-rior to doxorubicin or thiotepa for reducing

recur-rence of T1 and high-grade Ta tumors There is

no evidence that any intravesical therapy affects

the ultimate rate of progression to muscle-invasive

disease

Because there is risk of progression to invasive disease even after intravesical therapy,cystectomy may be considered as an initial treat-ment option in certain cases Among factors asso-ciated with increased risk of progression are largetumor size, high grade, tumor location in a sitepoorly accessible to complete resection, diffusedisease, infiltration of lymphatic or vascularspaces and prostatic urethral involvement Thisrecommendation is based on the panel's expertopinion

muscle-Recommendations for Index Patient No 3

A patient with CIS or high-grade T1 cancer who has had at least one course of intravesical therapy.

This recommendation is based upon panelopinion rather than evidence in the outcomes ta-bles However, other data in the literature doshow a substantial risk of progression to muscle-invasive cancer in patients with diffuse CIS andhigh-grade T1 tumors (Herr, 1997; Hudson andHerr, 1995; Kiemeney, Witjes, Heijbroek, et al.,1993) It is not certain whether intravesical thera-

py alters this risk of progression Thus, there aresome patients with symptomatic disease or high-grade tumors who may want to move directly tocystectomy Physicians should present specificinformation about the risks of cystectomy andmethods for urinary reconstruction to patientswho are contemplating bladder removal

Option: Adjuvant intravesical chemotherapy or

immunotherapy is an option for treatment after

endoscopic removal of low-grade Ta bladder

cancers

Guideline: Intravesical instillation of either

BCG or mitomycin C is recommended for

treatment of CIS and for treatment after

endo-scopic removal of T1 tumors and high-grade

Ta tumors

Option: Cystectomy may be considered for

initial therapy in some patients with CIS or T1

tumors

Option: Cystectomy may be considered as an

option for patients with CIS or high-grade T1cancers that have persisted or recurred after aninitial intravesical treatment

Option: Further intravesical therapy may be

considered as an option for patients with CIS

or high-grade T1 cancers that have persisted orrecurred after an initial intravesical treatment

Archived Document— For Reference Only

This recommendation is also based on the

pan-el's expert opinion Optimal dosing schemes for

intravesical chemotherapy and immunotherapy

have not been established, but weekly instillations

for at least six weeks are used most often There is

additional evidence showing that some patients

will respond to second induction regimens,

particu-larly with BCG Repeat intravesical therapy may

be appropriate in patients who develop a late rence after previous complete response to an in-travesical agent Data are insufficient, however, toallow conclusions about the role of drug combina-tion regimens or alternating therapies

recur-Archived Document— For Reference Only

To develop the recommendations in this Report

on the Management of Non-Muscle-Invasive

Bladder Cancer (Stages Ta, T1 and Tis), the AUA

Bladder Cancer Clinical Guidelines Panel used an

explicit approach (Eddy, 1992) This approach

at-tempts to arrive at practice policy

recommenda-tions through mechanisms that systematically take

into account relevant factors for making selections

between alternative interventions Such factors

in-clude estimation of the outcomes from the

inter-ventions, consideration of patient preferences and

assessment of the relative priority of the

interven-tions for a share of limited health care resources

when possible For estimating the outcomes of

in-terventions, emphasis is placed on the use of

sci-entific evidence When panel opinion is

neces-sary, the explicit approach calls for explaining

why it was necessary and/or for discussion of the

factors considered

In developing its recommendations, the panel

made an extensive effort to review the literature

on non-muscle-invasive bladder cancer and to

esti-mate outcomes of alternative treatment modalities

as accurately as possible The panel members

themselves served as proxies for patients in

con-sidering preferences with regard to health and

eco-nomic outcomes

The review of the evidence began with a

litera-ture search and extraction of data as described on

page 10 The data available in the literature were

displayed in evidence tables From these tables,

with reference back to the original studies when

necessary, the panel developed comparative

out-comes estimates for the following therapeutic

al-ternatives for treating non-muscle-invasive bladder

cancer:

• Transurethral resection of bladder tumor

(TURBT) without adjuvant therapy;

• TURBT with thiotepa;

• TURBT with doxorubicin;

• TURBT with mitomycin C;

• TURBT with bacillus Calmette-Guérin (BCG)

The panel utilized the FAST*PRO®

meta-analysis software described on pages 10-11 to

combine the outcomes evidence from the variousstudies The resulting probability estimates aredisplayed in the outcomes tables on pages 23-24.Because tumor removal is the standard of practice,there is no evidence comparing removal versusnonremoval Thus, this report concentrates oncomparison of the above listed adjuvant therapies

as used to reduce the likelihood of tumor rence and progression

recur-The panel generated its recommendationsbased on the probability estimates shown in theoutcomes tables and on expert opinion Theserecommendations were graded according to threelevels of flexibility as determined by strength ofevidence and the expected amount of variation inpatient preferences The three levels of flexibilityare defined as follows (Eddy, 1992):

1 Standard: A treatment policy is considered a

standard if the health and economic outcomes

of the alternative interventions are sufficientlywell-known to permit meaningful decisionsand there is virtual unanimity about which in-tervention is preferred

2 Guideline: A policy is considered a guideline

if 1) the health and economic outcomes of theinterventions are sufficiently well-known topermit meaningful decisions and 2) an appre-ciable but not unanimous majority agree onwhich intervention is preferred

3 Option: A policy is considered an option if 1)

the health and economic outcomes of the ventions are not sufficiently well-known topermit meaningful decisions, 2) preferencesamong the outcomes are not known, 3) pa-tients' preferences are divided among the alter-native interventions and/or 4) patients are in-different about the alternative interventions

inter-A standard has the least flexibility inter-A line has significantly more flexibility, and optionsare the most flexible In this report, the terms areused to indicate the strength of the recommenda-tions A recommendation was labeled a standard,for example, if the panel concluded that it should

guide-be followed by virtually all health care providersfor virtually all patients A guideline generally

Chapter 1: Methodology

Archived Document— For Reference Only

denotes a recommendation supported by objective

data but not with sufficient strength to warrant a

designation of standard An option in this report

would include treatments for which there appears

to be equal support in the literature or ones for

which there is insufficient published information

to support a stronger recommendation Also, as

noted in the above definition, options can exist

be-cause of insufficient evidence or bebe-cause patient

preferences are divided In the latter case

particu-larly, the panel considered it important to take into

account likely preferences of individual patients

with regard to health outcomes when selecting

from among alternative interventions

Literature searches were performed mostly

us-ing the MEDLINE®database Some articles

(in-cluding some that predate the inception of

MED-LINE®in 1966) were added for review based on

panel members' own knowledge The initial

MEDLINE®search was performed in 1989 A

subsequent literature search was performed in

1992 with the search terms bladder neoplasms and

carcinoma, transitional cell Several update

searches were also performed, the last in 1998

All searches were restricted to English-language

articles on human subjects Update searches were

further restricted to articles containing the term

superficial in the bibliographic record The

result-ing database of search results was used as the

ba-sis for selecting articles for data extraction No

attempt was made to weight the articles by quality

of data

The panel as a group reviewed the abstracts

and selected the relevant articles for data

extrac-tion From a database of 5,712 citations, the panel

selected 227 articles for retrieval and data

extrac-tion A comprehensive data-extraction form was

devised by the panel to capture as much pertinent

information as possible from each article

Appendix B contains a sample of this form

Reviewers were recruited from residencies at

the panel members' medical centers and trained to

use the review form The articles selected by the

panel were retrieved and divided among the

re-viewers, who extracted the articles and transcribed

the data onto the forms The panel rejected 46 ticles at this stage because of reasons such as in-adequate methods, no relevant data or superses-sion by a later article from the same source (seeAppendix A, Figure A-5) The result was 181 ar-ticles to be used as the source of data for this re-port All data were entered into a MicrosoftAccess 97®database (Microsoft, Redmond,Washington), then double-checked at a later time.The bar graph in Figure A-1 on page 50 cate-gorizes by year of publication the number of arti-cles reviewed and the number accepted for dataextraction The articles extracted range in year ofpublication from 1964 to 1998 The graph inFigure A-2 categorizes the articles by source The

ar-majority came from the Journal of Urology,

Urology and The British Journal of Urology.

The data resulting from the article-selectionand data-extraction process just described werecombined to generate the comparative probabilityestimates for alternative interventions displayed inthe outcomes tables (Tables 2-5) on pages 23-24

A variety of methods can be used to combine comes evidence from the literature to generateprobability estimates The methods chosen de-pend on the nature and quality of the evidence

out-For example, if there is only one good randomizedcontrolled trial (RCT), the results of that one trialalone may be used in the outcomes tables Otherstudies of lesser quality may be ignored

For non-muscle-invasive bladder tumors, datafrom more than 30 RCTs were available to gener-ate comparative estimates of recurrence and pro-gression probabilities for the different treatmentalternatives The small patient size of many ofthese studies, however, made them suboptimal forestimating probabilities of treatment complica-tions, particularly uncommon complications

Thus, the panel opted to include data from clinicalseries as well as from RCTs for generating com-plications estimates

If a number of studies have some degree of evance to a particular cell or cells in an outcomestable, meta-analytic methods may be used

rel-Different specific methods are available depending

on the nature of the evidence For this report, thepanel elected to use the Confidence ProfileMethod (Eddy 1989; Eddy, Hasselblad andShachter, 1990), utilizing FAST*PROTMcomputer

Literature search,

article selection and

Archived Document— For Reference Only

software (Eddy and Hasselblad, 1992) The

Confidence Profile Method allows analysis of data

both from RCTs and from single-arm studies that

are not controlled

Two different meta-analytic approaches were

used For estimates of recurrence and

progres-sion, data from multi-armed RCTs were combined

meta-analytically to determine the difference

be-tween two treatment alternatives All alternatives

were compared in pairs For estimates of

treat-ment complications, meta-analysis was performed

to combine data from single arms of more than

one study, including the relevant single arms of

multi-armed RCTs An estimate of the probability

of each complication was computed Thus, the

outcomes table for treatment complications (Table

5, page 24) shows the probabilities of a particular

complication occurring for each treatment choice

In the tables for recurrence and progression (page

23), the numbers represent the difference in

prob-ability between two treatment interventions For

example, if 10 percent of patients given

interven-tion A had a recurrence, and 13 percent of patients

given intervention B had a recurrence, the number

listed in the outcomes table would be 3 percent

Different study results may be caused by a

number of factors, including differences in patient

populations, how an intervention was performed

or the skill of those performing the intervention

Because of such differences, the meta-analyses all

used a random effects or hierarchical Bayesian

model in combining data from different studies

A random effects model assumes that for each

study there is an underlying true rate for the

out-come being assessed It further assumes that this

underlying rate varies from study to study, and

that the variation in the true rate is normally

dis-tributed

The probability distributions produced by the

Confidence Profile Method can be described using

a mean or median estimate of the probability and

a confidence interval In this report, the

95-per-cent confidence interval is such that the

probabili-ty (Bayesian) of the true value being outside the

interval is 5 percent The width of the confidence

interval indicates the degree of uncertainty about

the estimated probability, reflecting such factors as

differences in outcomes data being combined from

different studies and differences in the size of the

studies

The results presented in this document haveseveral limitations, mostly because of how dataare reported in the literature Determination ofoutcomes by tumor stage is an important example.Such determination was limited by the fact thatmost studies, including RCTs, categorize patientswith CIS, Ta or T1 tumors as all having "superfi-cial bladder cancer," without reporting stage-spe-cific results In addition, most studies do not seg-regate response by tumor grade Yet, stage andgrade, even in the absence of muscle invasion, arestrong predictors of recurrence and progression.These problems limited the ability of the panel tomake strong, evidence-supported recommenda-tions

A possible limitation in using RCTs for analysis is that, by their nature, most RCTs com-pare two or more different treatments, whereas thecomparison of all alternative treatments may bewhat is desired for purposes of evidence-basedclinical practice guidelines Ideally, this shouldnot be a problem because the two-way compar-isons should line up consistently in relation to oneanother, and it should be possible to generatenumbers that show with consistency the differ-ences among all of a given set of treatments

meta-For example, assume that a two-way son based on meta-analysis of RCTs shows the es-timated probability of patient survival to be 20percent greater after treatment A than after treat-ment B Another two-way comparison based onother RCTs shows this estimated probability to be

compari-10 percent greater after treatment B than aftertreatment C Therefore, a comparison of treat-ments A and C should logically show a difference

of 30 percent in favor of treatment A In reality,such consistency of results is uncommon whenbased on data from different studies Among thereasons for possible inconsistencies are differ-ences in study designs, differences in the mea-sures used, patient differences and simple randomvariation

With regard to intravesical drugs, studies differ

on dosages, strains, administration protocols andmixtures of patients Nevertheless, the panel ulti-mately decided to combine data from studies thatdiffer in these areas, because such differences

software (Eddy and Hasselblad, 1992) The

Confidence Profile Method allows analysis of data

both from RCTs and from single-arm studies that

are not controlled

Two different meta-analytic approaches were

used For estimates of recurrence and

progres-sion, data from multi-armed RCTs were combined

meta-analytically to determine the difference

be-tween two treatment alternatives All alternatives

were compared in pairs For estimates of

treat-ment complications, meta-analysis was performed

to combine data from single arms of more than

one study, including the relevant single arms of

multi-armed RCTs An estimate of the probability

of each complication was computed Thus, the

outcomes table for treatment complications (Table

5, page 24) shows the probabilities of a particular

complication occurring for each treatment choice

In the tables for recurrence and progression (page

23), the numbers represent the difference in

prob-ability between two treatment interventions For

example, if 10 percent of patients given

interven-tion A had a recurrence, and 13 percent of patients

given intervention B had a recurrence, the number

listed in the outcomes table would be 3 percent

Different study results may be caused by a

number of factors, including differences in patient

populations, how an intervention was performed

or the skill of those performing the intervention

Because of such differences, the meta-analyses all

used a random effects or hierarchical Bayesian

model in combining data from different studies

A random effects model assumes that for each

study there is an underlying true rate for the

out-come being assessed It further assumes that this

underlying rate varies from study to study, and

that the variation in the true rate is normally

dis-tributed

The probability distributions produced by the

Confidence Profile Method can be described using

a mean or median estimate of the probability and

a confidence interval In this report, the

95-per-cent confidence interval is such that the

probabili-ty (Bayesian) of the true value being outside the

interval is 5 percent The width of the confidence

interval indicates the degree of uncertainty about

the estimated probability, reflecting such factors as

differences in outcomes data being combined from

different studies and differences in the size of the

studies

The results presented in this document haveseveral limitations, mostly because of how dataare reported in the literature Determination ofoutcomes by tumor stage is an important example.Such determination was limited by the fact thatmost studies, including RCTs, categorize patientswith CIS, Ta or T1 tumors as all having "superfi-cial bladder cancer," without reporting stage-spe-cific results In addition, most studies do not seg-regate response by tumor grade Yet, stage andgrade, even in the absence of muscle invasion, arestrong predictors of recurrence and progression.These problems limited the ability of the panel tomake strong, evidence-supported recommenda-tions

A possible limitation in using RCTs for analysis is that, by their nature, most RCTs com-pare two or more different treatments, whereas thecomparison of all alternative treatments may bewhat is desired for purposes of evidence-basedclinical practice guidelines Ideally, this shouldnot be a problem because the two-way compar-isons should line up consistently in relation to oneanother, and it should be possible to generatenumbers that show with consistency the differ-ences among all of a given set of treatments

meta-For example, assume that a two-way son based on meta-analysis of RCTs shows the es-timated probability of patient survival to be 20percent greater after treatment A than after treat-ment B Another two-way comparison based onother RCTs shows this estimated probability to be

compari-10 percent greater after treatment B than aftertreatment C Therefore, a comparison of treat-ments A and C should logically show a difference

of 30 percent in favor of treatment A In reality,such consistency of results is uncommon whenbased on data from different studies Among thereasons for possible inconsistencies are differ-ences in study designs, differences in the mea-sures used, patient differences and simple randomvariation

With regard to intravesical drugs, studies differ

on dosages, strains, administration protocols andmixtures of patients Nevertheless, the panel ulti-mately decided to combine data from studies thatdiffer in these areas, because such differences

Limitations

Archived Document— For Reference Only

appear to have little or no effect on rates of

recur-rence and progression This is demonstrated by

the comparative data from BCG studies shown in

Table C-1 in Appendix C The panel also

com-bined data from studies that differed in follow-up

duration The panel did this because those studies

that had longer follow-up and showed actuarial

data indicated that differences in recurrence and

progression apparently develop early in the

fol-low-up period and remain relatively constant over

time

Differences in how authors define and record

complications are also a common problem Some

authors report the most minor of complications,

whereas other authors fail to report complications

at all If a complication is rare and the panel only

analyzes those papers that report the

complica-tion, there will be a significant overestimation of

its frequency The panel attempted to determineappropriate denominators for such complications,but the possibility of overestimation still exists.Another common difficulty is negative publi-cation bias Studies with poor results are lesslikely to be published because either they are nev-

er submitted for publication or they are rejectedlater For example, with regard to articles com-paring adjuvant therapies to TURBT alone, there

is the possibility that studies demonstrating tive results for adjuvant therapies versus TURBTalone may never have been published

nega-Notwithstanding such limitations, the paneldid find high-quality outcomes data available ontreatment of non-muscle-invasive bladder cancerand in sufficient quantity for computing reliableprobability estimates to aid physicians and pa-tients in their treatment decision making

Archived Document— For Reference Only

Chapter 2: Non-muscle-invasive bladder cancer

and its management

Chapter 2 provides an overview of the current

knowledge base for bladder carcinomas that have

not invaded the detrusor muscle (stages Ta, T1

and Tis) This overview includes summary

de-scriptions of known etiologic factors, tumor types

and characteristics, histology and natural history

Described also are current methods of diagnosis,

staging and grade classification, as well as

meth-ods for treating primary tumors and for predicting

likelihood of tumor recurrence and subsequent

progression

Malignant transformation of a normal bladder

cell begins with alteration of the cell's DNA

Epidemiologic evidence implicates chemical

car-cinogens as a causative agent in many patients

Specifically, the evidence points to aromatic

amines, such as 2-naphthylamine and

4-amino-biphenyl, which are present in cigarette smoke

and various industrial chemicals Cigarette smoke

metabolites, secreted into smokers' urine, have

been estimated to cause more than 50 percent of

the bladder cancer cases in the United States and

Europe (Bryan, 1993; Vineis, Martone and

Randone, 1995) Occupational exposure to

indus-trial chemicals has been estimated to account for

up to 20 percent of bladder cancer cases in the

United States, often with latency periods of more

than 30 years (Bryan, 1993; Cole, Hoover and

Friedell, 1972; Silverman, Levin, Hoover, et al.,

1989a, 1989b)

Other reported risk factors for bladder cancer

include phenacetin-related analgesic abuse; that

is, consumption in excessive quantities over a

10-year period (Piper, Tonascia and Metanoski,

1985) The average latency period is 22 years,

and most patients have renal pelvic tumors as well

(Johansson, Angervall, Bengtsson, et al., 1974)

The drug cylophosphamide, which is used to treat

a variety of conditions including Hodgkin's

dis-ease, can also lead to development of bladder

can-cer; the causative agent is believed to be

cy-clophosphamide's urinary metabolite acrolein

The latency period is a relatively short 6-13 years,and most of the tumors are high grade when diag-nosed (Cohen, Garland, St John, et al., 1992;

Fernandes, Manivel, Reddy, et al., 1996; O'Keane,1988; Talar-Williams, Hijazi, Walther, et al.,1996) Bladder cancer can result from radiothera-

py used to treat uterine or cervical cancer or otherpelvic tumors, and again the tumors tend to behigh grade when diagnosed (Quilty and Kerr,1987; Sella, Dexeus, Chong, et al., 1989)

Much of today's research regarding bladdercancer etiology focuses on the role of altered ge-netic mechanisms, some of which have been un-der study for years Oncogenes are an example.These are altered forms of alleles called proto-oncogenes Proto-oncogenes encode a growthfactor as well as receptor proteins and are neces-sary for normal cell development The alteredforms, oncogenes, lead to unregulated cell divi-sion and, in bladder cancer, increased recurrenceand progression (Del Senno, Maestri, Piva, et al.,1989; Tiniakos, Mellon, Anderson, et al., 1994).Tumor-suppressor genes such as p53 have alsobeen under intense study One of the functions ofp53 is to direct DNA-damaged cells toward apop-tosis before the DNA can replicate, thus suppress-ing proliferation of cells that have genetic abnor-malities (Harris and Hollstein, 1993) If p53 itself

is damaged, it may no longer be able to function(Cordon-Cardo, 1995) Bladder cancers associat-

ed with mutated p53 tend to behave quite sively (Cordon-Cardo, 1995; Esrig, Elmajian,Groshen, et al., 1994), and altered genes like p53may serve as markers to indicate the relative ag-gressiveness of individual tumors

aggres-More than 90 percent of all bladder cancers inthe United States and Europe, both muscle inva-sive and noninvasive, are transitional cell

Etiology

Major types of bladder cancer

Archived Document— For Reference Only

carcinomas originating in the urothelium that

forms the bladder lining (Fleshner, Herr, Stewart,

et al., 1996) Transitional cell carcinomas may

appear in a variety of configurations including

exophytic papillary tumors (the most common

configuration); flat patches of carcinoma in situ

(CIS); nodular tumors; sessile growths; and mixed

growths such as high-grade papillary tumors

to-gether with flat CIS

Less common types of bladder cancer include

squamous cell carcinoma, which accounts for 3 to

7 percent of all bladder cancers in the United

States, and adenocarcinoma, which accounts for

less than 2 percent of primary bladder cancers

(Kantor, Hartge, Hoover, et al., 1988; Koss, 1975;

Lynch and Cohen, 1995) Adenocarcinoma is the

most common type of cancer in exstrophic

blad-ders (Bennett, Wheatley and Walton, 1984;

Nielsen and Nielsen, 1983)

In the United States, squamous cell carcinoma

of the bladder is associated with factors such as

chronic urinary tract infections and chronic

irrita-tion from long-term indwelling catheters and

uri-nary calculi Spinal-cord injured patients, in

par-ticular, are at risk for squamous cell bladder

can-cer (Navon, Soliman, Khonsari, et al., 1997) In

North Africa, where squamous cell carcinoma is

also associated with bilharzial infection from

Schistosoma haematobium, this type of bladder

cancer is much more prevalent than in the United

States and Europe (Zhang and Steineck, 1997) In

Egypt, squamous cell carcinoma accounts for

more than 75 percent of all bladder cancers

(El-Bolkainy, Mokhtar, Ghoneim, et al., 1981)

The bladder has three main histologic layers:

1) the urothelium; 2) the suburothelial layer of

loose connective tissue called the lamina propria;

and 3) the detrusor muscle (muscularis propria)

The urothelium in the normal bladder consists of

three to seven layers of transitional-type epithelial

cells, which vary in appearance from cell layer to

cell layer Those in the luminal cell layer are

large, flat umbrella cells bound together by tight

junctions The urothelium's deepest layer is

com-posed of basal cells resting on and attached to an

extracellular-matrix basement membrane (basal

lamina) Between these cell layers are one to five

The muscularis mucosa, when parts of it are sent in small, fragmented tumor specimens ob-tained by biopsy or transurethral resection, may

pre-be difficult for the pathologist to distinguish fromdetrusor muscle The result may be overstaging

of a tumor erroneously thought to be sive, but actually limited to the lamina propria

muscle-inva-The detrusor muscle itself consists of interlacingbundles of smooth muscle with a covering of adi-pose tissue (or peritoneum)

Hematuria is the usual first sign of bladdercancer, present at least microscopically in almostall patients with cystoscopically detectable tumors(Messing and Valencourt, 1990) In some cases,bladder irritability accompanied by urgency, fre-quency and dysuria will be present in addition tohematuria This set of symptoms is associatedwith diffuse CIS or muscle-invasive disease

Routine diagnostic methods for bladder cancerinclude a thorough history, especially regardingexposure to known carcinogens; a physical exami-nation; urine analysis; and a cystoscopic examina-tion of the bladder and urethra Diagnostic cysto-scopies today are usually outpatient proceduresdone with a flexible or rigid cystoscope under lo-cal anesthesia Diagnostic tools available in addi-tion to cystoscopy include cytologic assessmentand several new urinary tests approved by theFDA for detection or monitoring of recurrences ofbladder cancer These diagnostic aids will be dis-cussed later

Also, because hematuria can originate where than in the bladder, imaging of the entireurinary tract may be warranted at the time ofhematuria evaluation (using a range of availableimaging techniques) Once a diagnosis of bladdercancer is established, upper tract imaging is usuallyperformed

else-Histology

Diagnosis

Archived Document— For Reference Only

Urinary cytology

Microscopic cytologic examination of bladder

cells, collected from voided urine or from bladder

washings (which potentially yield more cells), has

long been part of the diagnostic armamentarium

for bladder cancer Cytology is frequently used as

well in post-treatment follow-up to detect possible

cancer recurrence (Grégoire, Fradet, Meyer, et al.,

1997)

Cytologic assessment is least sensitive for

de-tecting low-grade bladder tumors Because the

cells of these well-differentiated tumors often

closely resemble normal urothelial cells,

malig-nant features may be difficult to recognize In

ad-dition, the cells are relatively cohesive and may

not shed into the urine in sufficient numbers for

analysis Cytologic detection rates for low-grade

tumors are generally less than 30 percent (Kannan

and Bose, 1993; Murphy, Soloway, Jukkola, et al.,

1984)

Cytology is much more effective for detecting

high-grade tumors, the cells of which have

sub-stantial abnormal characteristics Also, reduced

cellular cohesion in high-grade tumors produces

greater numbers of cells shed into the urine

Cytology is especially valuable for detecting CIS,

with detection rates as high as 90 percent

(Grégoire, Fradet, Meyer, et al., 1997; Messing

and Catalona, 1998)

Now being investigated are ways to improve

cytology performance such as through

immunos-taining with either Lewisxantigen or cytokeratin

20 (Klein, Zemer, Buchumensky, et al., 1998;

Pode, Golijanin, Sherman, et al., 1998)

Recently developed urine

assay tests

The bladder tumor antigen (BTA) test was the

first of several tests developed to detect bladder

cancer recurrence based on urinary presence of

certain antigens, nuclear matrix proteins or other

substances known to be associated with bladder

malignancies (Fradet, 1998; Grossman, 1998;

Sardosdy, White, Soloway, et al., 1995) The

original BTA test has been followed by two newer

assays, BTA statTM

and BTA TRAKTM

, which mayoffer improved performance over the original

(Ellis, Blumenstein, Ishak, et al., 1997; Sarosdy,

Hudson, Ellis, et al., 1997)

A nuclear matrix protein test, NMP-22, has

been approved by the FDA to detect occult or

rapidly recurring bladder cancer after a thral resection of a bladder tumor (TURBT)

transure-NMP-22 has been reported to be more sensitivebut less specific than cytology as a diagnostictool; however, experience with the test is still lim-ited (Stampfer, Carpinito, Rodriguez-Villanueva,

et al., 1998)

Another test is based on telomerase activity.Telomerase is an enzyme responsible for chromo-some-end (telomere) maintenance and is com-monly active in cancer The TRAP test, or telom-eric repeat amplification protocol assay, is report-

ed to be highly sensitive but may require specialcollection techniques (Kavaler, Landman, Chang,

et al., 1998; Linn, Lango, Halachmi, et al., 1997).These and other tests now under development(including easy-to-use point-of-service dip-stick-type assays) appear promising but need to be test-

ed themselves by time and experience to provetheir clinical utility (Bandalament, 1998;

Grossman, 1998; Johnston, Morales, Emerson, etal., 1997) Most of these tests are limited by in-sufficient specificity to allow treatment decisionsbased on a positive test, although they appear to

be more sensitive than cytology for detecting mors, especially low-grade lesions

tu-Diagnostic TURBT

TURBT (described on page 18) is the usualmethod for initial diagnosis of bladder cancer ATURBT is performed both to excise all visible tu-mors and to provide specimens for pathologicevaluation to determine tumor stage and grade(Shelfo, Brady and Soloway, 1997) Additionalloop or cold-cup biopsies may be taken to evalu-ate other areas of the urothelium, and a bimanualpalpation before and after resection may providefurther information on tumor size and depth ofpenetration A repeat TURBT may be performed

in cases of incompletely resected Ta and T1 mors

Laser coagulation may be used to ablate mors in some cases, especially recurrent lesions

tu-If laser therapy is used, cold-cup biopsies may betaken for pathologic evaluation

For staging of bladder cancer, the originalJewett-Strong system (1946), modified byStaging

Archived Document— For Reference Only

Marshall (1952, 1956), has generally given way to

the TNM (tumor, node, metastasis) system

devel-oped jointly by the American Committee on

Cancer Staging and the International Union

Against Cancer (Hermanek and Sobin, 1992;

Fleming, 1997) Depth of tumor penetration is

the crucial element in both systems Table 1

shows the TNM classifications for primary

tu-mors, adapted from the American Joint

Committee on Cancer (AJCC) staging manual

(Fleming, 1997)

Based on large data sets from hospitals in the

United States and Germany, 30-35 percent of

bladder cancer patients at time of diagnosis

pre-sent with stage Ta or Tis disease (confined to the

urothelium); 25-30 percent have stage T1 disease

(penetration into the lamina propria); and 30-35

percent present with muscle-invasive disease in

stage T2 or higher (Fischer, Waechter, Kraus, et

al., 1998; Fleshner, Herr, Stewart, et al., 1996)

Use of the term "superficial"

As noted in the introduction to this report

(page i), the three non-muscle-invasive stages––

Ta, T1 and Tis––are often grouped together under

the label "superficial." This is an imprecise label,

and the AUA Bladder Cancer Guidelines Panel

believes its use should be discouraged All three

stages are indeed superficial to the detrusor

mus-cle, but otherwise differ markedly from one

an-other, particularly with regard to tumor recurrenceand progression (Epstein, Amin, Reuter, et al.,1998; Fitzpatrick, 1993; Levi, La Vecchia,Randimbison, et al., 1993)

Low-grade Ta tumors, for example, tend to cur as low-grade Ta tumors Progression to mus-cle-invasive disease takes place in less than 10percent of patients with these tumors (Holmäng,Hedelin, Anderström, et al., 1997; Kiemeney,Witjes, Heijbroek, et al., 1993; Prout, Barton,Griffin, et al., 1992) By contrast, T1 tumors have

re-a progression rre-ate of 30-50 percent in spite oftherapy, and there is a high risk of death because

of progression (as high as 30 percent as reported

in studies with long-term patient follow-up) (Herr,1997; Kiemeney, Witjes, Heijbroek, et al., 1993).Treatment recommendations may be substantiallydifferent for patients with Ta tumors than for pa-tients with T1 disease

Basic characteristics of stages Ta, T1 and Tis

Stage Ta tumors are confined to the urothelium(above the basement membrane) and have a papil-lary configuration described by Johansson andCohen (1996) as resembling "seaweed" protrudinginto the lumen of the bladder Most Ta tumors arelow grade

Stage T1 tumors have penetrated below thebasement membrane and infiltrated the laminapropria, but have not gone so far as the detrusormuscle Most T1 tumors are papillary, but many

of those that have penetrated the deepest into thelamina propria are nodular (Heney, Nocks, Daly,

et al., 1982)

Hasui, Osada, Kitada, et al (1994) reportedsignificant increases in cancer recurrence and pro-gression when T1 tumors penetrated to the level

of the muscularis mucosa and beyond a two-foldincrease in incidence of recurrence and an eight-fold increase in incidence of progression Hasuiand colleagues, as well as other authors, have pro-posed subdividing the stage T1 classification intostages T1a and T1b, using the muscularis mucosa

as the dividing line Unfortunately, the laris mucosa makes an unreliable point of refer-ence, being discontinuous and often difficult toidentify; but there is growing evidence that tumorpenetration to this depth does have important im-plications, such as a higher risk of later detrusormuscle invasion (Epstein, Amin, Reuter, et al.,1998; Holmäng, Hedelin, Anderström, et al.,

muscu-Table 1:

Staging of primary bladder cancer tumors (T)

Ta: Noninvasive papillary carcinoma

Tis: CIS (anaplastic "flat tumor" confined to

urothelium)

T1: Tumor invades lamina propria

T2: Tumor invades muscularis propria

T2a: Invades superficial muscularis propria

T2b: Invades deep muscularis propria

T3: Tumor invades perivesical fat

T3a: Invades microscopic perivesical fat

T3b: Invades macroscopic perivesical fat

(extravesical mass)

T4: Tumor invades prostate, uterus, vagina,

pelvic wall or abdominal wall

T4a: Invades adjacent organs (uterus,

ovaries, prostate stoma)

T4b: Invades pelvic wall and/or abdominal

wall

Archived Document— For Reference Only

1997) In addition, tumor infiltration of vascular

or lymphatic spaces may portend a poor outcome

if unequivocally present, but it is often difficult to

determine with certainty

In a stage by itself, CIS (stage Tis) has been

defined as high-grade (anaplastic) carcinoma,

which, like stage Ta, is confined to the

urotheli-um, but with a flat, disordered, nonpapillary

con-figuration and a likelihood of being

underdiag-nosed (Epstein, Amin, Reuter, et al., 1998) CIS

can be focal, multifocal or diffuse On

cystoscop-ic examination, it usually appears as a slightly

raised, reddened patch of velvety mucosa—but

often it is endoscopically invisible Much about

this peculiar cancer is poorly understood,

includ-ing its natural history and the mechanism by

which it spreads The most ominous

characteris-tic of CIS may be mulcharacteris-ticentricity (Koch and

Smith, 1996) Even when the extent of spread

seems limited, additional areas frequently exist

not only in other parts of the bladder, but in the

distal ureters, the urethra or the prostatic ducts

Prostatic urethral biopsy may yield additional

in-formation in patients with diffuse CIS If

untreat-ed, CIS is likely to progress to muscle-invasive

disease in a substantial percentage of patients

(Hudson and Herr, 1995)

Numerous classification systems for grading

transitional cell carcinomas of the bladder have

been developed and published over the past few

decades Although no single system has yet

emerged to win universal acceptance, the most

widely used systems all share important

charac-teristics In particular, they all tend to group

blad-der carcinomas similarly into three principal

grades based mainly on degree of anaplasia

(Bergkvist, Ljungqvist and Moberger, 1965;

Epstein, Amin, Reuter, et al., 1998; Koss, 1975)

The three grades—low (grade 1), intermediate

(grade 2) and high (grade 3)—correspond

respec-tively to well differentiated, moderately

differenti-ated and poorly differentidifferenti-ated tumors Grade has

been shown to be a highly predictive indicator of

future tumor behavior with regard to both

recur-rence and progression

In addition to grade, the most clinically usefulfactors for predicting recurrence and progression

in patients who present with non-muscle-invasivedisease are tumor stage, tumor size, number of tu-mors and the presence of CIS Stage is particu-larly important As reported previously (page 16),the likelihood of progression to muscle-invasivedisease is much greater for T1 tumors than for Tatumors (Herr, 1997; Holmäng, Hedelin,

Anderström, et al., 1997; Kiemeney, Witjes,Heijbroek, et al., 1993; Prout, Barton, Griffin, etal., 1992) Moreover, the likelihood of muscle in-vasion appears to be greater still for T1 tumorsthat have penetrated beyond the muscularis mu-cosa (Epstein, Amin, Reuter, et al., 1998;

Holmäng, Hedelin, Anderström, et al., 1997)

Other possible predictors of recurrence and/orprogression include DNA ploidy, blood groupantigens and various molecular markers DNAploidy is measured by flow cytometry, often per-formed on tumor tissue embedded in paraffin

Diploid tumor cells tend to be low grade and lowstage, indicating a favorable prognosis, whereastetraploid and aneuploid tumors have a greaterlikelihood of progression and/or recurrence(Gustafson, Tribukait and Esposti, 1982a, 1982b).However, there is some question as to the inde-pendent prognostic value of ploidy over gradealone (Masters, Camplejohn, Parkinson, et al.,1989; Murphy, Chandler and Trafford, 1986;

Tachibana, Deguchi, Baba, et al., 1991, 1993)

Loss of cell-surface blood group antigens,such as ABO antigens and the Lewisxantigen, hasbeen associated with an unfavorable prognosis insome studies, but other studies have found no cor-relation with either recurrence or progression(Newman, Carlton and Johnson, 1980; Richie,Blute and Waisman, 1980; Yamada, Fukui,Kobayashi, et al., 1991)

Molecular markers, such as abnormal sion of the tumor-suppressor genes p53 and pRb(see page 13), show considerable promise asprognostic indicators for bladder cancer, especial-

expres-ly for predicting aggressive tumor behavior(Cordon-Cardo, 1998; Cote, Dunn, Chatterjee, etal., 1998; Grossman, Liebert, Antelo, et al., 1998).Reliable methods for clinical use of such markersare still being developed

Prognostic indicators

Grade classification

Archived Document— For Reference Only

In most cases of non-muscle-invasive bladder

cancer, tumors are treated initially with TURBT,

fulguration and/or laser therapy A careful

cysto-scopic examination of all bladder surfaces, the

urethra and the prostate precedes resection (Koch

and Smith, 1996) Findings with prognostic

sig-nificance are noted during this examination

Noted, for example, is the position of tumors with

reference to the bladder neck and ureteral orifices;

tumor configuration, such as whether tumors are

papillary or sessile; and estimates of the number

of tumors and their sizes Following resection of

all visible tumors, adjuvant intravesical

chemotherapy or intravesical immunotherapy is

commonly used to prevent recurrences (Messing

and Catalona, 1998)

Resection and fulguration

of bladder tumors

As stated on page 15, a TURBT has two main

purposes: 1) complete eradication of all visible

tu-mors; and 2) tissue resection for pathologic

evalu-ation to determine grade and stage Fulgurevalu-ation

may be used on small lesions, but tissue still needs

to be obtained to determine grade and stage at

time of initial presentation

Koch and Smith (1996), in describing surgical

technique, recommend use of a continuous flow

resectoscope, which allows a relatively constant

volume in the bladder and thus minimizes

move-ment of tumors away from and toward the

resecto-scope When a tumor is removed, a separate

biop-sy can be taken at the base with the resecting loop

After which, healthy-appearing muscle fibers

should be visible at the base Necrotic-appearing

tissue implies an invasive carcinoma The

pres-ence of fat implies a full-thickness bladder wall

defect

Laser therapy

The Nd:YAG laser has so far proven to be the

most versatile wavelength for treating bladder

can-cer, but other wavelengths also have been used

(Koch and Smith, 1996; Smith, 1986) Results are

comparable to electrocautery resection, with little

difference in the recurrence rate (Beisland and

Seland, 1986)

However, tissue samples need to be obtainedbeforehand by means of cold-cup biopsies to de-termine tumor grade Assessing depth of tumorpenetration to determine stage is more problematicwith laser therapy Appropriate patients for thistherapy have papillary, low-grade tumors and ahistory of low-grade, low-stage tumors (Koch andSmith, 1996)

Intravesical chemotherapy and immunotherapy

Intravesical chemotherapy or immunotherapy

is most often used as adjuvant treatment to preventtumor recurrence following the transurethral resec-tion of primary non-muscle-invasive bladder tu-mors, including possible recurrence because of ia-trogenic implantation of tumor cells Intravesicaltherapy is also used to treat known existing tumors

in cases of CIS, which frequently cannot be

treat-ed adequately by resection or fulguration because

of diffuse involvement Intravesical therapy hasbecome the preferred primary treatment for CIS(Messing and Catalona, 1998) In addition, anagent intended for prophylaxis may in fact betreating undetected existing disease Basic proper-ties of the chief intravesical agents currently avail-able are described later in this chapter Chapter 3contains an evidence-based comparative outcomesanalysis of most of these agents

Thiotepa

Introduced in 1961, thiotepa is the oldest andone of the least expensive of the intravesicaldrugs It is an alkylating agent that acts by cross-linking nucleic acid It is administered in dosesranging from 30 mg in 30 ml of sterile water orsaline to 60 mg in 60 ml of water or saline Thelower dose appears to be as effective as the higherone in a comparative study, but the concentrations

in both were the same (Koontz, Prout, Smith, etal., 1981) The usual regimen consists of six toeight weekly instillations followed by monthly in-stillations for one year Thiotepa's low molecularweight of 189 allows partial absorption throughthe urothelium, with possible systemic toxicity.There is a risk of myelosuppression especially atthe 60 mg dose level White cell and plateletcounts are obtained before each instillation, andtreatment is delayed if necessary

Treatment alternatives

Archived Document— For Reference Only

Mitomycin C

Mitomycin C is an antibiotic that works by

in-hibiting DNA synthesis Because of its

moderate-ly high molecular weight of 329, there are few

problems with transurothelial absorption, and

myelosuppression is rare Dosage varies from 20

mg to 60 mg per instillation; a commonly used

dose is 40 mg in 40 ml of saline or sterile water,

administered weekly for eight weeks followed by

monthly instillations for one year Mitomycin C

is a very expensive agent There is evidence that

multiple follow-up instillations are only slightly

more beneficial than just a single instillation

ad-ministered within 24 hours after TURBT (Tolley,

Parmar, Grigor, et al., 1996)

Doxorubicin (adriamycin) and epirubicin

Doxorubicin is an anthracycline antibiotic able

to bind to DNA and inhibit synthesis It is not

cell cycle specific, but appears to be most

cyto-toxic in the S phase Its molecular weight of 580

is high, and absorption and systemic toxicity are

extremely rare Doses vary widely, from 10 mg

to 100 mg, in instillation schedules that range

from three times a week to once a month

Epirubicin, a derivative of doxorubicin, is an

an-thracycline analog that appears to be somewhat

better tolerated than doxorubicin (Kurth, Vijgh,

Kate, et al., 1991) A study by the European

Organization for Research and Treatment of

Cancer supports the possibility that epirubicin

may be most effective against cystoscopically

in-visible tumors existing at the time of resection

(Oosterlinck, Kurth, Schroder, et al., 1993) As of

1999, epirubicin was not available in the United

States, and the panel did not formally analyze

outcomes data for this drug

Bacillus Calmette-Guérin (BCG)

BCG, a live attenuated strain of

Mycobacterium bovis, was first used as a

tubercu-losis vaccine Its now widespread use as

intraves-ical immunotherapy for management of

noninva-sive bladder cancer began in the 1970s, heralded

by the 1976 Morales study (Morales, Eidinger

and Bruce, 1976) BCG has since become a

first-line treatment for CIS and has been shown to be

effective as prophylaxis to prevent bladder cancer

recurrences following TURBT (Cookson and

Sarosdy, 1992; Coplen, Marcus, Myers, et al.,

1990; DeJager, Guinan, Lamm, et al., 1991; Herr,Schwalb, Zhang, et al., 1995; Lamm, Blumen-stein, Crawford, et al., 1995)

Several substrains of BCG have been used, cluding the Pasteur, Armand-Frappier, Tice,Evans, Tokyo, Dutch (RIVM) and Connaughtstrains Although all of them are derived fromone original strain developed at the PasteurInstitute, the viability of BCG organisms per mil-ligram of vaccine may vary with different sub-strains and from lot to lot within the same sub-strain (Cummings, Hargreave, Webb, et al., 1989;Kelley, Ratliff, Catalona, et al., 1985; Messingand Catalona, 1998)

in-BCG's mechanism of action is not fully stood, but it clearly involves a strong inflammato-

under-ry immunologic host response with release of terleukins and other cytokines (Morales, Eidingerand Bruce, 1976; Ratliff, Haaff and Catalona,1986) Most patients develop this inflammatoryimmunologic response during a typical course ofsix weekly instillations Some patients do not re-quire that many instillations; other patients re-quire more than six (Brosman, 1992; Eure,Cundiff and Shellhammer, 1992; Haaff, Dresner,Ratliff, et al., 1986) Optimal dosing and instilla-tion schedules have not yet been established Themost common side effects of BCG are cystitis andhematuria The most serious is BCG sepsis

in-BCG therapy is contraindicated in patients whoare immunocompromised, who have liver disease

or a history of tuberculosis Also, there is agreater potential for sepsis if the drug is adminis-tered too soon after TURBT or traumatic catheter-ization

Valrubicin and Interferon

Valrubicin and interferon are two newer agentsfor treating bladder cancer Valrubicin, a semi-synthetic analog of doxorubicin, has receivedFDA approval for treatment of patients with CISwho have not responded to BCG therapy and inwhom immediate cystectomy would be associatedwith unacceptable morbidity or mortality Thepanel did not formally review clinical studies us-ing valrubicin, but only about one of five patientswith BCG-refractory CIS was reported to havehad a complete response

Interferons have antiproliferative, genic and immunostimulatory properties; and re-combinant interferon alpha-2b has been effective

antiangio-Archived Document— For Reference Only

for treating CIS in some clinical trials (Glashan,

1990) Interferon side effects tend to be minimal

However, as prophylaxis for noninvasive bladder

cancer, interferons may not be as effective as

BCG (Kalble, Beer and Staehler, 1994) More

clinical studies are needed before this is

deter-mined Because not enough studies have been

published as yet, the panel did not have sufficient

outcomes evidence available to include interferon

in the analysis described in Chapter 3

Two key questions regarding follow-up after

treatment for noninvasive bladder cancer are: 1)

how long should follow-up be continued and 2)

at what intensity? The classic follow-up protocol

consists of cystoscopic and cytologic

examina-tions every three months for 18 to 24 months

af-ter the initial tumor, then every six months for the

following two years and then annually

(Fitzpatrick, 1993; Messing and Catalona, 1998)

The results of the first three-month cystoscopy

often predict the future recurrence pattern

(Fitzpatrick, West, Butler, et al., 1986) Less tense surveillance may be indicated if patientshad Ta papillary low-grade tumors that werecompletely resected and if the results of the firstthree-month cystoscopic examination were nega-tive (Abel, 1993; Gulliford, Burney and

in-Petruckitch, 1993)

Bladder cancer, however, may recur even afterlong, disease-free intervals, indicating a need forlifelong surveillance (Cookson, Herr, Zhang, etal., 1997; Messing and Catalona, 1998;

Thompson, Campbell, Kramer, et al., 1993)

Surveillance includes periodic upper-tract ing, especially if the treated cancer was CIS, oth-

imag-er high-grade tumors or tumors located near aureteral orifice (Cookson, Herr, Zhang, et al.,1997; Miller, Eure and Schellhammer, 1993)

The new urine assay tests described on page

15, such as BTA statTM, BTA TRAKTMand

NMP-22, may ultimately prove sensitive enough to place a portion of the routine follow-up cysto-scopic examinations

re-Follow-up

Archived Document— For Reference Only

The AUA Bladder Cancer Clinical Guidelines

Panel analyzed available outcomes data for both

the potential benefits and the potential

complica-tions of alternative approaches to treating

non-muscle-invasive bladder cancer Results of the

panel's analysis are summarized as probability

estimates in the comparative outcomes tables on

pages 23-24 of this chapter The data extraction

and evidence combination processes that

pro-duced the probability estimates are described

on pages 10-11 of Chapter 1 The raw data are

in a technical supplement to this report,

Evidence Working Papers, which is available

from the AUA

Types of outcomes include direct outcomes

that are experienced by the patient firsthand and

that affect the patient directly Some, such as

dy-suria or an increase in urinary frequency, may be

short-term side effects of treatment Others, such

as urinary frequency due to bladder contraction

from scarring, may occur on a continuing basis

long after treatment has been completed Direct

outcomes are often of great importance to the

pa-tient simply because they impact them By

con-trast, asymptomatic tumor recurrence, although an

extremely important outcome, may not

necessari-ly seem important to the patient because it is not

experienced directly From the patient's

stand-point, however, it is important as a proxy for

fu-ture adverse events such as the likelihood of

re-peated treatment and possibly more aggressive

treatments such as radical cystectomy

The outcomes tables (Tables 2-5) on pages

23-24 summarize results following Confidence

Profile (FAST*PROTM) meta-analyses of combined

outcomes data from the bladder cancer treatment

literature, as described in Chapter 1 Results are

displayed in the tables as probability estimates

In most cases, a 95-percent confidence interval is

reported along with a median estimate of the

probability It should be noted that "median" inthese tables is the median of the probability distri-bution resulting from FAST*PRO®meta-analysis(Eddy, Hasselblad and Shachter, 1990); it is notthe median of an array of individual study results.The G/P columns in Table 5 show (G) the number

of patient groups for a given outcome and (P) thetotal number of patients in those groups A cellmarked "No Data" indicates insufficient ex-tractable data for a given outcome

For assessing potential benefits and possibleadverse effects of interventions for treatment ofnon-muscle-invasive bladder cancer, the panel de-termined that the following outcomes are the mostimportant:

• Probability of tumor recurrence;

• Risk for tumor progression; and

• Complications of treatment

Recurrence

Non-muscle-invasive bladder cancer has a highrate of new tumor formation within the bladder.New tumor formations following treatment aregenerally termed "recurrences" (although in manycases they might be more accurately termed "re-occurrences") Recurrent tumors may have pro-gressed to a higher grade and/or higher stage, butmost recurrences of non-muscle-invasive bladdercancer (especially of initially diagnosed grade 1,stage Ta disease) are not muscle invasive and car-

ry relatively little risk of metastasis or death

However, recurrent disease will likely require arepetition of therapy

For the patient and physician assessing ment alternatives, the probability of recurrencefollowing treatment is obviously an important out-come to consider The first two outcomes tables,Tables 2 and 3 on page 23, display comparativedifferences in the estimated probability of recur-rence following use of intravesical agents as adju-vant therapy after transurethral resection of blad-der tumor (TURBT) The comparative differences

treat-Chapter 3: Outcomes analysis for treatments of

non-muscle-invasive bladder cancer

The outcomes tables

Archived Document— For Reference Only

in Table 2 are statistically significant; those in

Table 3 are not

Each intravesical agent in the tables is

com-pared to TURBT alone and to each of the other

agents, based on data from multi-armed

random-ized controlled trials (RCTs) All of the agents,

when used as adjuvant therapy after TURBT, are

shown to result in a lower estimated probability of

recurrence compared to use of TURBT alone

Some of the differences in Table 2 are quite large

An example is the use of TURBT plus BCG

ver-sus the use of TURBT alone With a median

dif-ference of 30 percent (95% CI 19-39 percent), the

table shows a significantly lower estimated

proba-bility of recurrence in favor of TURBT plus BCG

(All median differences in Tables 2-4 are in favor

of the first intervention listed in each pair.)

The last column of Table 2 displays another

way of showing the results "Number needed to

treat" (NNT) is the number of patients who must

be treated with the superior treatment of each pair

in order to reduce by one the estimated number of

patients with recurrences For example, in

"TURBT plus doxorubicin versus TURBT alone,"

the NNT is 10 Thus, if 10 patients are treated

with the superior treatment of this pair, TURBT

plus doxorubicin, and another 10 patients with

TURBT alone, one less recurrence can be

expect-ed in the patient group receiving TURBT plus

doxorubicin

The recurrence tables are further discussed on

pages 25-26 of this chapter The basic

methodol-ogy is described on page 11 of Chapter 1

Progression

The risk for progression of

non-muscle-invasive bladder cancer, depends on a number of

factors, including size and number of tumors

and, especially, their grade and stage

Low-grade, stage Ta tumors, although they have a

tendency to recur, have a low risk for

progres-sion to muscle invaprogres-sion, whereas progresprogres-sion

rates of 30-50 percent have been reported for

high-grade, stage T1 disease (Herr, 1997)

Most of the available studies reporting results

of treatment do not adequately address the risk for

long-term progression, notwithstanding the

impor-tance of this outcome Consequently, the results

displayed in Table 4 (page 23) are based on fewer

RCTs than were available for calculating the

re-sults shown in Table 2-3 (page 23)

Complications

of treatment

After review and evaluation of all possible verse outcomes from treatment for non-muscle-in-vasive bladder cancer, the panel grouped theseoutcomes into three general categories: 1) Localbladder symptoms; 2) Systemic symptoms; and 3)Other Their estimated probabilities for occur-rence with particular treatments are displayed inTable 5 on page 24

ad-Local bladder symptoms are the most commonadverse outcomes experienced by patients under-going treatment for non-muscle-invasive bladdercancer The most frequently observed immediatesymptoms are irritative lower urinary tract prob-lems including dysuria, frequency/nocturia, ur-gency, pain and cramping and passing of debris inthe urine, including blood or clots Patients mayalso experience bacterial cystitis, urinary inconti-nence or bladder perforation Long-term adverseoutcomes related to local bladder symptoms in-clude bladder contracture

Although local bladder symptoms can be vere, systemic symptoms are more threatening

se-These include flu-like symptoms such as gia, fever, chills and malaise Systemic infectiouscomplications may also result from treatment

arthral-Among these problems are pulmonary or hepaticchanges, pneumonia, pneumonitis, hepatitis, epi-didymitis, prostatitis and urethral infection Also,patients might experience nausea, vomiting, rash

or indirect adverse outcomes such as pression

myelosup-Some patients might experience adverse comes severe enough that the intervention must beinterrupted or even stopped entirely In rare cir-cumstances, deaths have been reported with inter-ventions used for non-muscle-invasive bladdercancer, although the frequency of death is notgenerally recorded

out-Cost, inconvenience and quality-of-life issues

Cost, inconvenience and quality-of-life issuesare certainly important to patients Few meaning-ful data are readily available from the literature toaddress these issues for non-muscle-invasive blad-der cancer, and for the most part, these are areas

in need of future research Nevertheless, in regard

to cost, panel members knew from their own periences that there are major differences between

ex-Archived Document— For Reference Only

Archived Document— For Reference Only

Archived Document— For Reference Only

intravesical agents, usually with mitomycin C

be-ing by far the most expensive Table 6 shows

comparative 1999 patient charges from a

universi-ty-based hospital setting of one of the panel

mem-bers Generalization is not possible, however, as

costs of intravesical agents may vary substantially

both regionally and from institution to institution

There are many retrospective studies reporting

outcomes of interventions for non-muscle-invasive

bladder cancer However, RCTs are the most

use-ful for analysis, eliminating many of the problems

encountered with uncontrolled or poorly

con-trolled studies Accordingly, the panel decided to

extract data only from RCTs for comparison of

treatments with regard to lowering the probability

of recurrence and progression

Yet, even with the advantage of having RCTs

available for analysis, the panel found the

evalua-tion process complicated for the following

rea-sons:

1 The patients in the different studies varied by

stage, grade, prior treatments and tumor

histo-ry, presence or absence of CIS and length of

time with bladder cancer The patient mix

was generally undifferentiated, and the studies

did not break down outcomes relative to

pa-tient characteristics

2 The treatment protocols differed in the number

and timing of medication instillations, doses

and frequency However, even where different

dosing regimens were compared within or

be-tween studies, no significant outcome

differ-ences were noted for different treatment

regi-mens (see Table C-1 in Appendix C)

3 The outcomes measured were reported in ous ways Recurrence rates were reportedvariably as number of patients with recur-rences, number of recurrences, number of oc-currences per unit of time or time of first re-currence For progression, some studies re-ported patients with different types ofprogression without indicating whether therewas overlap among these patients

vari-4 The follow-up times varied significantly, bothwithin and between studies

5 In some studies, patients unsuccessful withone treatment were crossed over to take othertreatments

Despite these limitations, the panel found theanalysis useful One important conclusion thepanel drew from the data evaluation was thattreatment-related reductions in the probability ofrecurrence were detected within the first year ofobservation in most studies Furthermore, the ob-served reductions were carried forward into sub-sequent years; that is, differences observed withinthe first year remained stable over time

Thiotepa after TURBT was compared withTURBT alone in 10 RCTs Meta-analysis ofthese studies demonstrated that thiotepa afterTURBT lowers the probability of recurrence morethan TURBT alone (Table 2, page 23) The medi-

an estimate of the difference is approximately 19percent (95% CI 9-28 percent) However, com-parative evaluation of these two treatments withregard to progression (Table 4, page 23) showed

no difference between them

The administration of BCG after TURBT wascompared with TURBT alone in six studies Asnoted previously (page 22), the meta-analysisyielded a 30 percent reduction in the probability

of recurrence using the BCG intervention (95%

CI 19-39 percent) For progression (Table 4, page23), the two treatments were compared in threestudies, yielding an estimated difference of 8 per-cent in favor of BCG (95% CI -0.2-15 percent).However, with the confidence interval overlappingzero, this difference is not statistically significant

Mitomycin C administration after TURBT

yielded a reduction in recurrence probability of

approximately 15 percent (95% CI 7-22 percent)

over TURBT alone For preventing progression,

there was no benefit to mitomycin C

administra-tion after TURBT versus TURBT alone

The administration of doxorubicin after

TURBT yielded a reduction of 10 percent in

prob-ability of recurrence (95% CI 2-17 percent) over

TURBT alone There was no evident reduction in

probability of progression using doxorubicin after

TURBT versus TURBT alone

Three studies compared the administration of

BCG versus thiotepa, each after TURBT These

studies showed a clear benefit to BCG over

thiotepa with a median difference in recurrence

probability of 30 percent (95% CI 10-47 percent)

Doxorubicin was also compared with BCG,

each after TURBT, in four studies The advantage

went to BCG over doxorubicin in reduction of

re-currence probability, with a median difference of

23 percent (95% CI 13-32 percent)

Other intervention comparisons (Table 3, page

23) of mitomycin C versus thiotepa, doxorubicin

versus thiotepa, mitomycin C versus BCG and

mitomycin C versus doxorubicin yielded no

statis-tically significant advantage for one treatment

over another in reducing the probability of

recur-rence

A comparative assessment of the probability of

progression reduction with the various treatment

modalities showed no evident differences among

them in the RCTs (Table 4, page 23)

The probability estimates for complications

from each of the interventions are displayed in

Table 5 on page 24 The table shows median

esti-mates of the probability of the complications

oc-curring, along with 95-percent confidence

inter-vals The complications data were obtained from

the randomized controlled trials as well as from

the non-randomized trials The panel decided

that, for complications, the data reported in all

studies were of value in assessing frequency of

complications for a given intervention, regardless

of whether the patients were involved in RCTs

A major difficulty in listing complications

stems from the familiar problem of how data are

summarized in the literature Some studies

classi-fy complications such as frequency and nocturiaseparately, whereas other studies combine themunder single headings Likewise, cutoff points forsignificant fever vary from study to study, andsome studies do not include a definition of fever.Thus, the data analysis required some judgment,and the combinations of data required some arbi-trary assignments by the data extractors and thepanel The panel specifically addressed outliersand made decisions whether to accept the data atface value or to exclude these data in situationswhere they could not be accurately or usefullycategorized

Some specific complications are not included

in all studies This might have been because theynever occurred or because they were never tracked

or recorded, depending upon the protocol ture Such arbitrary deletions in the studies canaffect the overall analysis of the complications.However, there is no way to account for these dif-ferences in reporting It is important to recognizethat some complications do not apply to each ofthe treatment modalities, and that each interven-tion has its own unique spectrum of complications

struc-as well struc-as complications in common with otherinterventions For example, almost all of the in-terventions induce some problem with local blad-der symptoms, but only BCG is likely to induceevidence of systemic infection

The panel decided to delete certain tions entirely from the analysis Among thesewere granulomas, bladder mucosal erythema andhypotension Such complications were deletedbecause they had no clinical significance to thepatient in the short or long term, or were so un-usual that they would not affect treatment deci-sions by patients or recommendations by thephysician The complication of urethral stricturewas deleted from the complications listing be-cause it was reported in so few patients

complica-Other complications were combined to easeinterpretation An example of this is epididymitisbeing included together with prostatitis and ure-thral infections

Certain review articles were of value for viding data about complications However, sucharticles were examined carefully to ensure thatthere was no duplication of data from other stud-ies in the review that would overweight certaincomplications

pro-Outcomes summary:

treatment complications

Archived Document— For Reference Only

Death associated with any of the interventions

for non-muscle-invasive bladder cancer is

ex-tremely unusual There are certainly risks

associ-ated with the anesthesia and surgical procedure

for TURBT Overall, however, in the panel's

ex-pert opinion, the risk of death would be much less

than 1 percent for each of the interventions The

panel believes that such a low rate is not likely to

be a significant factor in the selection of therapy

by a patient or in its recommendation by a

physi-cian

Local bladder effects occur with the greatest

frequency and are direct outcomes felt by

pa-tients Thus, they are the adverse outcomes most

likely to govern a patient's decision to select oneintervention over another Systemic complica-tions affect a minority of patients, yet can be quiteprofound with BCG therapy The risk of BCGsepsis may be a factor that influences patient pref-erences or physician recommendations

In general, the balance between likely cial outcomes (such as reduced recurrence rates)and likely adverse outcomes (such as local com-plications) will dictate which intervention a pa-tient will select for treatment of non-muscle-inva-sive bladder cancer

benefi-Archived Document— For Reference Only

The AUA Bladder Cancer Clinical Guidelines

Panel generated the recommendations in this

chapter based upon analysis of comparative

out-comes data from both randomized controlled

tri-als (RCTs) and clinical series and upon expert

opinion The recommendations apply to

treat-ment of patients with non-muscle-invasive,

transi-tional cell carcinoma of the bladder, including

CIS (CIS) as well as stages Ta and T1 tumors

The panel evaluated comparative data for the

fol-lowing treatment methods in particular:

• Transurethral resection of bladder tumor

(TURBT);

• TURBT plus thiotepa;

• TURBT plus doxorubicin;

• TURBT plus mitomycin C;

• TURBT plus bacillus Calmette-Guérin (BCG)

As explained in Chapter 1 (page 9), panel

rec-ommendations were graded according to three

levels of flexibility as determined by strength of

evidence and the expected amount of variation in

patient preferences The definitions of these three

levels of flexibility are repeated as follows from

Chapter 1:

1 Standard: A treatment policy is considered a

standard if the health and economic outcomes

of the alternative interventions are sufficiently

well-known to permit meaningful decisions

and there is virtual unanimity about which

in-tervention is preferred

2 Guideline: A policy is considered a guideline

if 1) the health and economic outcomes of the

interventions are sufficiently well-known to

permit meaningful decisions and 2) an

appre-ciable but not unanimous majority agree on

which intervention is preferred

3 Option: A policy is considered an option if 1)

the health and economic outcomes of the terventions are not sufficiently well-known topermit meaningful decisions, 2) preferencesamong the outcomes are not known, 3) pa-tients' preferences are divided among the alter-native interventions and/or 4) patients are in-different about the alternative interventions

in-A standard has the least flexibility in-A line has significantly more flexibility, and optionsare the most flexible In this report, the terms areused to indicate the strength of the recommenda-tions A recommendation was labeled a standard,for example, if the panel concluded that it should

guide-be followed by virtually all health care providersfor virtually all patients A guideline generallydenotes a recommendation supported by objectivedata but not with sufficient strength to warrant adesignation of standard An option in this reportwould include treatments for which there appears

to be equal support in the literature or ones forwhich there is insufficient published information

to support a stronger recommendation Also, asnoted in the above definition, options can exist be-cause of insufficient evidence or because patientpreferences are divided In the latter case particu-larly, the panel considered it important to take in-

to account likely preferences of individual tients with regard to health outcomes when select-ing from among alternative interventions

pa-The specific types of patients to whom thepanel's recommendations apply are termed indexpatients In recognition of the differences in deci-sion making that occur depending upon patientcircumstances, the panel defined three differentindex patients:

Index Patient No 1: A patient who presents

with an abnormal growth on the urothelium, butwho has not yet been diagnosed with bladder cancer;

Index Patient No 2: A patient with

estab-lished bladder cancer of any grade, stage Ta or

Chapter 4: Recommendations for management of

non-muscle-invasive bladder cancer

Treatment policies:

levels of flexibility

Index patients

Archived Document— For Reference Only