Medication adherence, molecular monitoring, and clinical outcomes in patients with chronic myelogenous leukemia in a large HMO Reina Haque*, Jiaxiao Shi, Joanie Chung, Xiaoqing Xu, Chant
Trang 1Medication adherence, molecular monitoring, and clinical
outcomes in patients with chronic myelogenous leukemia
in a large HMO
Reina Haque*, Jiaxiao Shi, Joanie Chung, Xiaoqing Xu, Chantal Avila,
Christopher Campbell, Syed A Ahmed, Lei Chen, Joanne E Schottinger
a r t i c l e i n f o
Article history:
Received 11 July 2016
Accepted 10 January 2017
a b s t r a c t
Objective: Our objective was to examine the association between adherence to tyrosine kinase inhibitors (TKIs) and molecular monitoring and the risk of disease progression or mortality among patients with chronic phase chronic myeloid leukemia (CML)
Design: We assembled a retrospective cohort of patients with CML (chronic phase, no prior cancer history, and confirmed to be Philadelphia chromosome positive) using data from electronic health records and chart reviews Medication possession ratio (MPR) was used to measure drug adherence
Setting: A large, community-based, integrated health plan in Southern California
Participants: The cohort consisted of 245 adult patients (18 years old) with Philadelphia positive chronic phase CML diagnosed from 2001 to 2012 and followed through 2013 Main outcome measures: In survival analyses, we examined the association of TKI adher-ence (MPR) and polymerase chain reaction (PCR) monitoring test frequency with the composite clinical outcome, progression to accelerated phase diseaseeblast crisis or mortality (progression-free survival) The cohort was followed for a maximum of 13 years (median 4.6 years)
Results: Over 90% of the cohort initiated TKI therapy within 3 months of diagnosis, and the mean MPR was 88% (SD 18%) Virtually all patients (96%) started on imatinib The rates of progression to accelerated phaseeblast crisis and mortality were lower in patients with greater TKI adherence (20.4/1000 years) versus lower adherence (27.0/1000 person-years) Patients who underwent PCR monitoring had a significantly reduced risk of progres-sion or mortality, which was seen in patients with high and low TKI adherence status from both the groups (hazard ratio [HR] 0.07 [95% CI 0.03-0.19 if MPR>90%] and HR 0.70 [95% CI 0.02-0.21 if MPR<90%])
Conclusion: Our results suggest that close clinical monitoring, which includes PCR monitoring
in patients with high and low TKI drug adherence, is associated with a lower risk of pro-gression or mortality
© 2017 American Pharmacists Association® Published by Elsevier Inc This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Introduction Chronic myeloid leukemia (CML) represents 14% of all leukemia diagnosed in the United States, with an incidence of 1.6 cases per 100,000 people per year Based on the National Cancer Institute's Surveillance, Epidemiology and End Results (SEER) Cancer Statistics Review, greater than 6660 cases of CML were diagnosed in 2015, and approximately 1140 people died from the disease that year.1 Survival dramatically improved in the last decade mainly because of the U.S Food
Disclosures: Research funding from Novartis and AstraZeneca on unrelated
topics has been provided to Kaiser Permanente Southern California
(Reina Haque) Lei Chen is an employee of Novartis.
Funding: Research funding to Kaiser Permanente Southern California was
supported by Novartis.
Previous presentation: The results of this study have been reported at a
poster presentation at the American Society for Hematology meeting, San
Francisco, CA, December 6, 2014.
* Correspondence: Reina Haque, PhD, MPH, Kaiser Permanente Research,
Department of Research and Evaluation, 100 S Los Robles, 2nd Floor, Pasadena,
CA 91101.
E-mail address: Reina.Haque@kp.org (R Haque).
Contents lists available atScienceDirect Journal of the American Pharmacists Association
j o u r n a l h o m e p a g e : w w w j a p h a o r g
http://dx.doi.org/10.1016/j.japh.2017.01.004
1544-3191/© 2017 American Pharmacists Association ® Published by Elsevier Inc This is an open access article under the CC BY-NC-ND license ( http://
Trang 2and Drug Administration (FDA) approval of thefirst
commer-cially available tyrosine kinase inhibitor (TKI), imatinib, in
2001 Before the availability of this life-saving drug, the
number of CML deaths in 2001 (n¼ 23,000) was markedly
higher Second-generation TKIs, such as dasatinib and
niloti-nib, were approved by FDA in the mid-2000s, but are not as
widely used as imatinib.2CML is a hematopoietic disorder of
pluripotent stem cells commonly associated with the distinct
presence of the Philadelphia chromosome The Philadelphia
chromosome results from a reciprocal translocation between
the long arms of chromosomes 9 and 22, t(9;22).3This
trans-location results in a fused BCR-ABL gene and the production of
a deregulated, constitutively active tyrosine kinase protein
The BCR-ABL oncogene is integral to the pathogenesis of CML,
and it leads to the increased malignant cell proliferation,
diminished cell differentiation, and decreased apoptosis The
development of TKIs to block unregulated activity through the
BCR-ABL tyrosine kinase pathway represents a breakthrough
in CML treatment.3
Via selective suppression of the BCR-ABL tyrosine kinase
pathway activity, TKIs such as imatinib, reduce the BCR-ABL
oncogene, inhibit leukemic cell progression, and decrease
mortality as evidenced by a 5-year overall survival of
approximately 90% in patients with CML.4The contrast is stark
in a comparison with the previous 5-year overall survival of
68% in patients with CML treated with interferon plus
cytarabine.3
The optimal duration of TKI therapy remains uncertain,
and few studies have examined whether the level of adherence
affects survival Current National Comprehensive Cancer
Network (NCCN) guidelines recommend that patients continue
to take TKIs indefinitely.5 Moreover, a few studies have
examined the overall survival by patient adherence levels, or
whether survival is affected by molecular monitoring.6-10It is
well known that medication adherence is a crucial factor in
achieving optimal CML outcomes, but few studies have
assessed the effect of regular molecular monitoring on
out-comes Monitoring response to TKI treatment based on
mo-lecular testing is one of the most important management
strategies for this disease After oral TKI treatment commences,
repeated molecular testing determines how well the TKI treatment is working, and whether fewer cells with the Phil-adelphia chromosomes exist Molecular monitoring tests are done using 2 methods: cytogenetics (karyotyping and fluo-rescence in situ hybridization) and polymerase chain reaction (PCR) In recent years, clinicians have shifted to using PCR for monitoring rather than cytogenetics Cytogenetics tests of circulating blood or aspirated bone marrow cells are used to identify the BCR-ABL gene and its characteristics (e.g., arrangement, number, size, shape) PCR is a more sensitive DNA test to detect the BCR-ABL fusion gene and other molec-ular abnormalities in circulating blood.5 With cytogenetic testing, complete response is demonstrated when no cells with the Philadelphia chromosome are found With PCR testing, a complete response is demonstrated when the BCR-ABL gene is not found in the circulating blood cells The frequency of these
2 tests ranges from every 3-6 months up to 18 months after CML diagnosis If response is poor to either of the monitoring tests, the dose of TKI is increased, or the patient is switched to another TKI In patients with even worse molecular results, the clinicians may recommend chemotherapy (interferon), stem cell transplantation, or palliative radiation One retrospective chart review study showed that regular molecular monitoring was associated with reduced risk of progression and higher likelihood of progression-free survival,11and another study showed that lack of monitoring was associated with increased risk of mortality.12As CML management has rapidly changed in the last decade, the combination of pharmacoepidemiologic and molecular monitoring data are critically needed to ulti-mately inform the optimal treatment course
Objective The objective of this study was to describe practice trends and clinical outcomes in a large, community-based, integrated health plan serving>4.2 million members in Southern Cali-fornia, known as Kaiser Permanente Southern California (KPSC) Specifically, this study examined the association of TKIs adherence and molecular monitoring on disease pro-gression and mortality
Methods Design, subjects, and setting
We assembled a retrospective cohort of patients with chronic phase CML, with confirmed Philadelphia-positive chromosome determined by chart review Patients were identified from the KPSC health plan's National Cancer Institute SEER-affiliated cancer registry; those with a prior history of cancer as determined by the registry were excluded The diagnosis of CML was determined by identi-fying patients with International Classification of Diseases for Oncology, Third Edition (ICD-O-3) codes for CML (ICD-O-3 9863) and CML, BCR/ABL positive (ICD-O-3 9875) from the KPSC cancer registry The cohort consisted of all 245 adult patients (18 years old) with Philadelphia-positive chronic phase CML diagnosed between January 1, 2001, and December 31, 2012 Subjects were followed through pro-gression to accelerated phase or blast crisis, health plan
Key Points
Background:
Chronic myeloid leukemia management has changed
rapidly over the past decade, and few studies have
examined the overall survival by patient adherence
levels or whether survival is affected by cytogenetic or
molecular monitoring
Findings:
Patients who underwent polymerase chain reaction
monitoring had a significantly reduced risk of
progression or mortality
Along with tyrosine kinase inhibitor medication
adherence, clinical monitoring with polymerase
chain reaction tests enhances survival
Trang 3disenrollment, death, or the end of the study (December 31,
2013), whichever occurredfirst We conducted chart reviews
for all patients to assess the incorrect coding of CML diagnosis
and phase We confirmed initial cases of chronic phase CML
by medical chart review of molecular testing results (only 2
cases were excluded because they were diagnosed at
accel-erated or blast phase) This study was approved by the KPSC
Institutional Review Board, which waived the requirement
for written informed consent
Data elements
Data were collected from electronic health records,
pharmacy dispensing records, and chart reviews including
demographics, TKI use (imatinib, nilotinib, dasatinib) and
adherence, molecular testing (usingfluorescence in situ
hy-bridization [FISH] or PCR assays), health care utilization, and
clinical outcomes The main outcomes examined were
pro-gression to accelerated phase or blast crisis (from chart
review) and mortality (captured by data linkage with the
State of California Master File of death certificates) The main
independent variables were TKI adherence, if response to
therapy was monitored by molecular testing, and a
cross-product term of these 2 variables Adherence was estimated
using the medication possession ratio (MPR), a commonly
used measure to assess medication compliance.13 MPR for
any TKI use was calculated as follows: numerator (total days
supply [excluding the last refill]) divided by the denominator
(last dispense date minusfirst dispense date).14,15MPR was
stratified into 2 groups, MPR greater than or equal to 90%
(indicating high adherence) and MPR less than 90%
(indi-cating low adherence) This cutoff was selected for
adher-ence, as studies have shown favorable clinical outcomes with
adherence levels greater than 90%.6-9,16For the 2 types of
molecular monitoring tests, cytogenetic tests (FISH assays)
were summarized for every 6-month period, and PCR test
utilization was summarized for every 3-month period These
intervals were selected according to the National
Compre-hensive Cancer Network guidelines.17 Reasons for TKI
discontinuation were ascertained from medical charts
Statistical analysis
The distribution (percent, frequency) of demographic
characteristics and health care utilization by patients were
mainly compared by year of CML diagnosis Overall
person-year rates of these events were stratified by MPR
Kaplan-Meier curves were used to examine survival by MPR; the
differences were tested using the log-rank test Adjusted
hazard ratios (HRs) and 95% CIs for the risk of the composite
clinical outcome (progression to accelerated phase [AP]eblast
crisis [BC] or mortality) were estimated using Cox proportional
hazards regression models The main independent variables in
the multivariate Cox proportional hazards regression analysis
included TKI MPR and PCR monitoring (these 2 variables were
not correlated, chi-square statistic; P¼ 0.46) Thus, collinearity
is not an issue for our analysis Because all patients who
un-derwent cytogenetic testing also unun-derwent PCR monitoring
(the more sensitive test), and because PCR monitoring is
becoming more commonly used in recent years, we examined
only PCR use in the Cox model The proportional hazards
assumption was checked using Schoenfeld residuals; no de-viations were observed We initially examined TKI use as a binary time-varying variable (0 up to initiation date; 1, there-after), but the results were similar to the HRs when we included only TKI MPR in the model Therefore, we presented the parsimonious model that included only the TKI MPR var-iable The results were similar because 94% of patients initiated TKI treatment within 3 months of CML diagnosis (in fact, 86%
of the subjects initiated TKI treatment within 30 days;Table 2) Because molecular testing by PCR is more sensitive than cy-togenetic methods (i.e., karyotyping, FISH assays), and given that the uses of these 2 tests were highly correlated, thefinal Cox model examined only PCR utilization Thefinal Cox model was run on the full cohort, and it included an interaction term for the TKI MPR status and PCR testing use Covariates that we examined included patient age, sex, race or ethnicity, and comorbidities based on the Charlson Comorbidity Index (CCI)
We used the Deyo adaptation of Charlson Combordity Index; data for the CCI was captured 1 year before initiation of CML therapy.18
Results Demographics
A total of 245 patients were newly diagnosed with chronic phase CML from 2001 to 2012 and treated with TKIs Of the
245 patients, 107 were female patients (43.7%) The cohort included 58.0% (n ¼ 142) non-Hispanic white patients and
Table 1 Demographic characteristics of patients with chronic phase CML diagnosed during 2001-2012
n (%) Year of CML diagnosis
Age at diagnosis (y)
Sex
Race Non-Hispanic white 142 (58.0)
Asian, Pacific Islander 23 (9.4) Others, mixed, unknown 8 (3.3) CCI (1 y before TKI therapy)
MPR (on any TKI)
Abbreviations used: CML, chronic myeloid leukemia; CCI, Charlson Comor-bidity Index; TKI, tyrosine kinase inhibitor; MPR, medication possession ratio.
Trang 422.4% (n¼ 55) Hispanic patients (Table 1) Half (49%) of the
subjects had a CCI greater than 1 Most of the patients received
a diagnosis before 65 years of age (77%)
TKI use and health care utilization
Table 2shows the distribution of TKI use and health care
utilization by diagnosis years (N¼ 245) Subjects were
strati-fied by year of CML diagnosis We grouped subjects into 3
periods: 2001-2006 (before molecular monitoring became
standard), 2007-2009 (when molecular monitoring became
more commonly used), and 2010-2012 (when
second-generation TKIs became available on the KPSC formulary)
More than 90% of the cohort initiated TKI therapy within 3
months of diagnosis, and mean MPR was 88.1% (SD¼ 17.6%)
Approximately two-thirds of these patients (64%, N¼ 157) had
an MPR greater than or equal to 90% for TKI use The mean
duration on TKI was 4.4 years (SD¼ 2.9 years) The majority of
patients initiated TKI within 30 days after the diagnosis of CML
(n¼ 210, 85.7%) Among TKI users, virtually all patients (96%)
started with imatinib Dasatinib and nilotinib had not been
used asfirst-line TKIs until 2010-2012 Of the 59 cases
diag-nosed from 2010 to 2012, only 10 and 4 patients used dasatinib
and nilotinib, respectively, as first-line therapy Half of the cohort (n ¼ 107) discontinued first-line treatment (mainly imatinib) Of these 107 patients, 73 (29.8%) subsequently switched to the newer-generation TKI drugs Thus, in the full cohort of 245 patients, 73 continued TKI therapy until the end
of follow-up, while 172 remained discontinued Reasons for discontinuation included general adverse effects (e.g., skin rash, muscle cramps; n ¼ 35); cardiovascular, lung, or abdominal effects (n¼ 6); incomplete cytogenetic or molec-ular response (n¼ 38); drug nonresponse (as mentioned in the clinical notes), provider recommended, or not compliant (n¼ 21); or bone marrow or stem cell transplant (n¼ 7)
During the follow-up period, 156 (63.7%) of the 245 patients were hospitalized for any reason at least once Patients were hospitalized approximately twice across all diagnosis years regardless of the type of TKI used asfirst-line therapy (Table 2) The ICD-9 code 205.10 (chronic myeloid leukemia without mention of remission) was the most frequent primary reason for hospitalization (11% of hospitali-zations), although there were several other reasons (e.g., pneumonia) However, hospitalization duration varied from 4 to 14 days for patients who initiated with imatinib in 2001-2006 and 2007-2009, respectively The hospitalization
Table 2
Distribution of TKI use and health care utilization
CML diagnosis, y 2001e2006 2007e2009 2010e2012
First-line TKI Imatinib Imatinib Imatinib Dasatinib Nilotinib Total
Total 112 (45.7) 74 (30) 45 (18.4) 10 (4.1) 4 (1.6) 245 (100) TKI use (not mutually exclusive)
Imatinib 112 (100) 74 (100) 45 (100) 3 (30.0) 1 (25.0) 235 (95.9) Dasatinib 26 (23.2) 25 (33.8) 12 (26.7) 10 (100) 0 (0.0) 73 (29.8) Nilotinib 8 (7.1) 6 (8.1) 6 (13.3) 2 (20.0) 4 (100) 26 (10.6) Any switching TKIs after discontinuation of first-line TKI (1 TKIs)
Yes 28 (25.0) 24 (32.4) 16 (35.6) 4 (40.0) 1 (25.0) 73 (29.8)
No 84 (75.0) 50 (67.6) 29 (64.4) 6 (60.0) 3 (75.0) 172 (70.2) Time to initiate TKI (d to first prescription after CML diagnosis)
0e7 20 (17.9) 32 (43.2) 12 (26.7) 6 (60.0) 1 (25.0) 71 (29.0) 8e14 43 (38.4) 26 (35.1) 19 (42.2) 2 (20.0) 2 (50.0) 92 (37.6) 15e30 26 (23.2) 10 (13.5) 10 (22.2) 1 (10.0) 0 (0.0) 47 (19.2)
Duration of TKI therapy (d on any TKI)
Mean, SD 2107.3, 1248.1 1491.9, 678.4 898.6, 358.1 761.5, 354.4 817.5, 533.1 1623.4, 1062.2 Discontinued first-line TKI use (chart review)
Yes 54 (48.2) 31 (41.9) 17 (37.8) 4 (40.0) 1 (25.0) 107 (43.7)
No 55 (49.1) 43 (58.1) 26 (57.8) 5 (50.0) 3 (75.0) 132 (53.9) Unknown 3 (2.7) 0 (0.0) 2 (4.4) 1 (10.0) 0 (0.0) 6 (2.4) Reasons for discontinuation (chart review)
Adverse events NOS, abdominal
effects, muscle cramps, skin rash
15 (27.8) 10 (32.3) 8 (47.1) 2 (50.0) 0 (0) 35 (32.7) BMT, stem cell transplant, 7 (13.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 7 (6.5) CVD, lung effects 2 (3.7) 3 (9.7) 0 (0.0) 1 (25.0) 0 (0.0) 6 (5.6) CyR, MR or non-response 20 (37.0) 10 (32.3) 6 (35.3) 1 (25.0) 1 (100) 38 (35.5) Drug nonresponse, non-compliant,
MD recommends drug switch
10 (18.5) 8 (25.8) 3 (17.6) 0 (0.0) 0 (0.0) 21 (19.6) Hospitalization duration (in d) among patients hospitalized
Median (min, max) 14 (2, 129) 4 (1, 127) 27 (2, 221) 15 (2, 29) 6 (5, 6) 10 (1, 221) Annualized hospitalizations among those hospitalized (per patient/per y)
Median (min, max) 0.44 (0.08, 6.17) 0.34 (0.14, 2.99) 0.62 (0.25, 6.42) 0.89 (0.57, 4.21) 0.61 (0.55, 68) 0.45 (0.08, 6.42) Annualized outpatient visits (per patient/per y)
Median (min, max) 20 (4, 98) 13 (3, 53) 11 (3, 43) 18 (7, 43) 17 (4, 30) 16 (3, 98)
Trang 5duration was 5.5 days for those who initiated with nilotinib,
and 15 days for dasatinib Patients who used imatinib as
first-line therapy during 2001-2006 had a higher median number
of outpatient visits per patient per year compared with those
receiving a diagnosis in 2007 or later
Outcomes
In the cohort of 245 patients, we observed that 13 patients
(5%) progressed to accelerated phase (AP) or blast crisis (BC); 8
(3.3%) died from CML; 21 (8.6%) died from other causes; 42
(17.1%) disenrolled from the health plan; and 161 (66%)
reached the study's end (December 31, 2013;Table 3) during
the 13-year study period (median, 4.6 years) Overall, the
median time to progression for the 13 patients who
pro-gressed to accelerated phase or blast crisis was 2.9 years
(range, 2.4 months to 8.4 years) Among 29 patients who died
during follow-up, the median time to mortality was 2.8 years
(range, 2.4 months to 12.2 years) Although hematopoietic
stem cell transplant was not examined as a study endpoint in
the survival analysis, 11 unique patients underwent this
pro-cedure (data not shown)
Table 4presents the person-year rates of progression to AP/
BC, disenrollment, death, or study's end, whichever occurred
first, stratified by MPR status The person-year rates of these
events correlated with the follow-up time, given most of these
events occurred among those diagnosed from 2001 to 2006
The person-year rate of progression to AP/BC was lower in
patients with MPR greater than or equal to 90% (7.2/1000
years) than with MPR less than 90% (15.8/1000
person-years) Patients with MPR90% (indicating high medication adherence) had a mortality rate of 20.4/1000 person-years versus 27.0/1000 person-years for patients with MPR<90% Cytogenetic testing (utilization summarized for every 6 months) use generally decreased over time in the 3 subcohorts (stratified by year of CML diagnosis: 2001-2006, 2007-2009, 2010-2012; Supplementary Figure 1A; Appendix), whereas percentage of patients who underwent PCR monitoring (sum-marized every 3 months) generally increased (Supplementary Figure 1B; Appendix) As illustrated in the graphs, Supplementary Figures 1A and B are consistent with the changing National Comprehensive Cancer Network (NCCN) guidelines, as molecular monitoring by PCR testing is becoming more common than cytogenetic testing in recent years.17 Supplementary Figure 2 (Appendix) displays the Kaplan-Meier curves comparing the mortality or progression to AP/BC rates among patients by PCR testing (3 tests per year during follow-up;<3 tests; no PCR testing), stratified by the adher-ence status (N¼ 245, TKI users) The 3 PCR monitoring tests per year reflects patients who had close to quarterly testing (ie, 3-4 tests per year) As demonstrated by the Kaplan-Meier plots, patients with more PCR monitoring tests had longer survival than those without PCR monitoring, independent of TKI adherence (P<0.01 for all graphs, log-rank tests); this supports the NCCN guidelines for more frequent molecular monitoring tests
In multivariable Cox models, we examined the association
of TKI adherence (MPR) and PCR monitoring test frequency with the composite clinical outcome (progression to AP/BC or mortality; Table 5 This composite outcome reflects
Table 3
Outcomes by TKI use status and year of CML diagnosis a
CML diagnosis, y 2001e2006 2007e2009 2010e2012 a
First-line TKI used Imatinib Imatinib Imatinib a Dasatinib Nilotinib Total
Total 112 (43.6) 74 (28.8) 45 (17.5) 10 (3.9) 4 (1.6) 245 (100) Vital status as of December 31, 2013
Disenrollment 20 (17.9) 17 (23.0) 4 (8.9) 1 (10.0) d 42 (17.1) Alive, reached end of study 59 (52.7) 50 (67.6) 39 (86.7) 9 (90.0) 4 (100) 161 (65.7) Follow-up duration (y)
Median (min, max) 7.4 (0.4, 12.9) 4.7 (0.3, 7.0) 2.9 (0.3, 3.9) 2.7 (1.1, 3.4) 3.2 (2.6, 3.6) 4.6 (0.2, 12.9) Time to disease progression (to AP/BC date)
Median (min, max) 4.9 (0.5, 8.4) 0.6 (0.3, 2.3) d d d 2.91 (0.19, 8.36) Time to mortality
Median (min, max) 3.4 (0.8, 12.2) 3.3 (1.5, 3.7) 0.5 (0.3,0.7) d d 2.8 (0.2, 12.2) Abbreviations used: AP, accelerated phase; BC, blast crisis.
a Cells are blank because no events were observed.
Trang 6progression-free survival We did notfind differences in the
risk of the composite clinical outcomes between having 1-2
and 3-4 PCR tests per year during the follow-up period,
possibly because of small sample size Therefore, the adjusted
model focused on the comparison between“PCR tests versus
no test” and the TKI MPR variable Table 5 gives the
com-parisons between TKI MPR and PCR test, adjusted for age at
cancer diagnosis, sex, race or ethnicity, and the CCI We
determined that the patients in both of the MPR groups who
underwent PCR monitoring had a significantly reduced risk of
the composite outcome (progression or mortality)dfor MPR
greater than or equal to 90%, HR¼ 0.07 (95% CI 0.03-0.19), P
<0.0001; for MPR less than 90% HR ¼ 0.07 (95% CI 0.02-0.21),
P<0.0001 Although the association between MPR and the
composite clinical outcome was not statistically significant,
we did observe that those with MPR greater than or equal to
90% had a reduced risk of progression or mortality (HR¼ 0.73
[95% CI 0.29-1.83] with PCR monitoring, and HR¼ 0.71 [95%
CI 0.25-2.02] without PCR monitoring) Interestingly, Hispanic
patients were 3.7-fold more likely to experience worse
out-comes than whites (HR¼ 3.68 [95% CI 1.51-8.96])
Discussion
In the treatment of chronic phase CML, orally administered
oncolytic drugs and TKIs are the cornerstone of CML
treat-ment In a retrospective cohort of patients with CML
(diagnosed with chronic phase disease) followed a maximum
of 13 years (median 4.6 years), we observed that patients who underwent molecular monitoring tests with PCR to assess disease response had a significantly reduced risk of the com-posite outcome (progression or mortality), which was seen in patients with both high (MPR90%) and low TKI adherence (MPR<90%) status This result was confirmed in the multi-variable Cox models by showing the reduced adjusted HRs for progression or mortality in both groups of patients with high and low TKI adherence status (after stratifying by yes/no PCR tests) The better survival among those who underwent any PCR molecular monitoring might have been attributed to the fact that such patients might have had their TKI dose increased, had more clinical management, or both We also determined that Hispanic patients had greater progression or mortality, possibly because of less follow-up care and management
Similarities exist between these data and published liter-ature For example, although patients had full medical insur-ance coverage including pharmacy benefits, the distribution of MPR was consistent with other United States' study pop-ulations.16 The mean MPR in this study was 88% in all the patients combined, and approximately one-third of these patients were in the lower MPR group (MPR<90%) In another retrospective study measuring adherence in India, Ganesan
et al.4determined that nonadherence was found in 29.6% of the study population Thefinding that molecular monitoring
Table 4
Person-year rates (95% CI), whichever came first, by medication possession ratio status (progression-free survival rates)
Variable MPR <90% MPR 90% Rate ratio (MPR 90/MPR <90%) Progression to AP/BC
Rate per 1000 PY 15.8 (6.3e32.5) 7.2 (2.6e15.7) 0.46 (0.15e1.36)
Disenrollment
Rate per 1000 PY 38.3 (22.3e61.3) 30.0 (19.4e44.3) 0.78 (0.42e1.45)
Death (all causes)
Rate per 1000 PY 27.0 (14.0e47.2) 20.4 (11.9e32.6) 0.75 (0.36e1.58)
Alive/Reach end of study
Rate per 1000 PY 117.2 (87.5e153.7) 130.7 (107.3e157.7) 1.12 (0.81e1.55)
Abbreviations used: MPR, medication possession ratio; AP, accelerated phase; BC, blast crisis; PY, person-year.
Table 5
Adjusted hazard ratio for the composite outcome (progression, mortality) a
TKI adherence (MPR)
and PCR test status
Contrasts Multivariate Cox model
Adjusted HR a 95% CI P value
Every 5 year increase in age 1.20 1.04e1.37 0.01 African American vs white 1.60 0.34e7.47 0.55
Abbreviations used: HR, hazard ratio; MPR, medication possession ratio; PCR, polymerase chain reaction; CCI, Charlson Comorbidity Index.
a Adjusted for age at cancer diagnosis, sex, race or ethnicity, and CCI.
Trang 7is associated with lower risk of progression or mortality is also
consistent with prior studies.11,12
The advantages of this study include the use of a cohort with
equal access to care, availability of comprehensive pharmacy
records, and the ability to ascertain outcomes and covariate
information from electronic health records Importantly, we
were able to examine clinical outcomes (progression and
mortality), in contrast to prior studies that mainly examined
molecular response.5-12,16,19,20As all KPSC patients had
phar-macy medication coverage, we could study outcomes without
the confounding effects of variable insurance coverage For
example, a previous case series report from Brazil showed that
patients with much lower adherence (because of lack of access
to drugs) correlated with a low cytogenetic and molecular
response.11
Limitations
This study has certain limitations Given the small sample
size, caution is warranted in interpreting the results;
how-ever, our study is still larger than some previously reported in
the literature.6-8The retrospective nature of this study relied
on chart reviews, which might have missed data or might
have included inconsistencies in documentation; however,
unlike other database studies, we had access to each patient's
electronic medical record Although diagnostic procedures
and NCCN recommendations have changed throughout the
13 years of this study, long-term follow-up is required to
assess survival As demonstrated inSupplementary Figures 1
and 2, and consistent with the changing NCCN guidelines,
molecular monitoring by PCR testing is becoming more
common than cytogenetic testing in recent years In addition,
estimating adherence with MPR relies on the assumption
that an electronic refill history is an accurate predictor of how
the patients are taking their medications This infers that
patients who pick up their medications in the pharmacy do
not forget to take their medications as prescribed However,
published studies demonstrate the use of different measures
of adherence (e.g., refill history, questionnaire, tablet
count-ing, microelectronic monitoring device).13-15 Generally,
one-third of patients are identified as nonadherent
Furthermore, CML treatment and monitoring guidelines
recommend quarterly PCR testing; however, patients or
cli-nicians might not have always followed that schedule
Nonetheless, having any number of PCR monitoring tests
after patients initiated treatment with a TKI was associated
with reduced risk of disease progression
Conclusion
As TKIs are the cornerstone for CML treatment, our study
highlights the need for pharmacists and clinicians to work
together to enhance molecular monitoring of treatment
out-comes to enhance survival Specifically, PCR monitoring
testing should be used to monitor disease progression closely
and to serve as a cue to increase TKI doses Currently, KPSC's
integrated health care system implemented an electronic
adherence tool that allows health care professionals to
pro-vide outreach calls to remind patients of overdue oral
medi-cations for breast cancer Combining the results of this study
and other published studies has motivated leaders of the KPSC integrated health care system to implement electronic
refill reminders for TKIs and clinician reminders for PCR monitoring tests
Acknowledgment Research funding for this study was supported by Novartis
We thank Lakshmi Kasthurirangan, PhD, of Novartis Health-care for editorial assistance with this manuscript
References
1 National Cancer Institute SEER cancer statistics review, 1975-2011 Available at: http://seer.cancer.gov/archive/csr/1975_2011/ Accessed November 1, 2015.
2 O'Brien S, Berman E, Moore JO, et al NCCN Task Force report: tyrosine kinase inhibitor therapy selection in the management of patients with chronic myelogenous leukemia J Natl Compr Canc Netw 2011;9(Suppl 2): S1eS25
3 Druker BJ, Guilhot F, O'Brien SG, et al, IRIS Investigators Five-year
follow-up of patients receiving imatinib for chronic myeloid leukemia N Engl J Med 2006;355(23):2408e2417
4 Ganesan P, Sagar TG, Dubashi B, et al Nonadherence to imatinib adversely affects event free survival in chronic phase chronic myeloid leukemia Am J Hematol 2011;86(6):471e474
5 Bisen A, Claxton DF Tyrosine kinase targeted treatment of chronic my-elogenous leukemia and other myeloproliferative neoplasms Adv Exp Med Biol 2013;779:179e196
6 Marin D, Bazeos A, Mahon FX, et al Adherence is the critical factor for achieving molecular responses in patients with chronic myeloid leuke-mia who achieve complete cytogenetic responses on imatinib J Clin Oncol 2010;28:2381e2388
7 Eliasson L, Clifford S, Barber N, Marin D Exploring chronic myeloid leu-kemia patients' reasons for not adhering to the oral anticancer drug imatinib as prescribed Leuk Res 2011;35:626e630
8 Noens L, van Lierde MA, De Bock R, et al Prevalence, determinants, and outcomes of nonadherence to imatinib therapy in patients with chronic myeloid leukemia: the ADAGIO study Blood 2009;113(22):5401e5411
9 Ibrahim AR, Eliasson L, Apperley JF, et al Poor adherence is the main reason for loss of CCyR and imatinib failure for chronic myeloid leukemia patients on long-term therapy Blood 2011;117:3733e3736
10 Wu EQ, Johnson S, Beaulieu N, et al Healthcare resource utilization and costs associated with non-adherence to imatinib treatment in chronic myeloid leukemia patients Curr Med Res Opin 2010;26:61e69
11 Goldberg SL, Chen L, Guerin A, et al Association between molecular monitoring and long-term outcomes in chronic myelogenous leukemia patients treated with first line imatinib Curr Med Res Opin 2013;29: 1075e1082
12 Chen C, Chen L, Dhanda R, et al Types, frequency, and adherence of molecular tests in chronic myelogenous leukemia at chronic phase (CML-CP) in the US community setting J Clin Oncol 2013;31(15) (Suppl): 7096
13 Sikka R, Xia F, Aubert RE Estimating medication persistency using administrative claims data Am J Manag Care 2005;11(7):449e457
14 Karve S, Cleves MA, Helm M, Hudson TJ, West DS, Martin BC Prospective validation of eight different adherence measures for use with adminis-trative claims data among patients with schizophrenia Value Health 2009;12(6):989e995
15 Martin BC, Wiley-Exley EK, Richards S, Domino ME, Carey TS, Sleath BL Contrasting measures of adherence with simple drug use, medication switching, and therapeutic duplication Ann Pharmacother 2009;43(1): 36e44
16 Darkow T, Henk HJ, Thomas SK, et al Treatment interruptions and non-adherence with imatinib and associated healthcare costs: a retrospective analysis among managed care patients with chronic myelogenous leukaemia Pharmacoeconomics 2007;25(6):481e496
17 National Comprehensive Cancer Network Tests and treatment responses
in chronic phase CML Available at: http://education.nccn.org/sites/ education.nccn.org/files/HEM-N-0333-0415%20CML%20Pt%20Vignette% 20Tool.pdf Accessed November 29, 2016.
18 Cleves MA, Sanchez N, Draheim M Calculating Charlson comorbidity index when analyzing short-term mortality using administrative data.
J Clin Epidemiol 1997;50(8):903e908
Trang 819 Peterson AM, Nau DP, Cramer JA, Benner J, Gwadry-Sridhar F, Nichol M.
A checklist for medication compliance and persistence studies using
retrospective databases Value Health 2007;10(1):3e12
20 Cid DM, Magalh~aes SM, Quixada AT, et al Chronic myeloid leukemia: an
overview of the determinants of effectiveness and therapeutic response
in the first decade of treatment with imatinib mesylate in a Brazilian
hospital Rev Bras Hematol Hemoter 2013;35(6):389e394
Reina Haque, PhD, MPH, Research Scientist II, Department of Research and
Evaluation, Kaiser Permanente Southern California, Pasadena, CA
Jiaxiao Shi, PhD, Research Scientist I, Department of Research and Evaluation,
Kaiser Permanente Southern California, Pasadena, CA
Joanie Chung, MPH, SAS/Analytical Programmer III, Department of Research and
Evaluation, Kaiser Permanente Southern California, Pasadena, CA
Xiaoqing Xu, MPH, Research Associate II, Department of Research and Evalua-tion, Kaiser Permanente Southern California, Pasadena, CA
Chantal Avila, MA, Project Manager, Department of Research and Evaluation, Kaiser Permanente Southern California, Pasadena, CA
Christopher Campbell, PharmD, Clinical Pharmacist, Department of Research and Evaluation, Kaiser Permanente Southern California, Pasadena, CA Syed A Ahmed, MD, Molecular Pathologist/Geneticist, Department of Research and Evaluation, Kaiser Permanente Southern California, Pasadena, CA Lei Chen, MD, Health Economics and Outcomes Researcher, Novartis Pharma-ceuticals Corporation, East Hanover, NJ
Joanne E Schottinger, MD, Medical Oncologist, Department of Research and Evaluation, Kaiser Permanente Southern California, Pasadena, CA
Trang 9Supplementary Figure 1 Percent of patients who underwent (A) cytogenetic (over 6-month intervals) and (B) PCR molecular monitoring tests (over 3-month intervals).
Trang 10Supplementary Figure 2 Survival probability for mortality or progression to accelerated phase/blast crisis in all patients, stratified by numbers of PCR testing (A) All patients (B) Survival probability among those with MPR <90% (C) Survival probability among those with MPR >90%.