Trends in Childhood Cancer Incidence in the U.S. (1992–2004) pdf

Surveil-lance, Epidemiology, and End Results SEER program were evaluated to assess incidence and trends of common primary cancers diagnosed between 1992 and 2004 among children aged birt

Trends in Childhood Cancer Incidence in the U.S.

(1992–2004)

Research, Department of Pediatrics, University of

Minnesota, Minneapolis, Minnesota.

Minnea-polis, Minnesota.

Exami-nation of population-based incidence data provides insight regarding etiology among various demographic groups and may result in new hypotheses The objective of the current study was to present updated information regarding childhood cancer incidence and trends in the U.S overall and among demo-graphic subgroups, including Asian/Pacific Islanders and Hispanics, for whom to the authors’ knowledge trends have not been previously examined.

Surveil-lance, Epidemiology, and End Results (SEER) program were evaluated to assess incidence and trends of common primary cancers diagnosed between 1992 and

2004 among children aged birth to 19 years Frequencies, age-adjusted incidence rates, and joinpoint regression results, including annual percent change (APC) in incidence rates (and 95% confidence intervals [95% CI]), were calculated.

pedi-atric cancer diagnoses combined There was a suggestion of an increase in

were observed for astrocytoma Rate increases were noted for hepatoblastoma (APC, 4.3%; 95% CI, 0.2%–8.7%) and melanoma (APC, 2.8%; 95% CI, 0.5%–5.1%) Differences by demographic group (sex, age, and race/ethnicity) are also described.

population-based surveillance and further etiologic studies Cancer 2008;112:416–32.

2007 American Cancer Society.

KEYWORDS: epidemiology, childhood cancer, incidence, trends

underscor-ing a need to monitor incidence rates The last comprehensive reports concerning U.S incidence trends included data through

Cancer Institute’s (NCI’s) annual Cancer Statistics Review (available

data regarding recent childhood cancer incidence and trends, including an analysis of trends in several demographic groups

MATERIALS AND METHODS

Data were obtained from the NCI’s Surveillance, Epidemiology, and

Address for reprints: Julie A Ross, PhD,

Depart-ment of Pediatrics, University of Minnesota, 420

Delaware Street SE, MMC 422, Minneapolis, MN

55455; Fax: (612) 626-4842; E-mail: rossx014@

umn.edu

Supported by National Institutes of Health Grant

Research Fund.

Received December 27, 2006; revision received

August 14, 2007; accepted August 16, 2007.

DOI 10.1002/cncr.23169

Published online 11 December 2007 in Wiley InterScience (www.interscience.wiley.com).

population Between 1992 and 2004, SEER actively

collected data on all cancer cases (excluding

nonme-lanoma skin cancers) in Connecticut, Hawaii, Iowa,

New Mexico, and Utah; in the metropolitan areas of

Atlanta, Detroit, Los Angeles, San Francisco-Oakland,

San Jose-Monterey, and Seattle-Puget Sound; and

from rural Georgia and the Alaskan Native Tumor

Registry The use of the expanded SEER dataset from

1992 onward permitted the evaluation of the most

recent trends and facilitated the calculation of

inci-dence rates among Hispanics and Asian/Pacific

Islanders

Histology and topography codes from the third

edition of the International Classification of Diseases

cases of 15 common cancers and 8 subtypes (Table

1) We included all first malignancies diagnosed

dur-ing the period 1992 through 2004 among those aged

19 years; 95% of diagnoses were confirmed by

his-tology Annual population estimates used in the

cal-culation of incidence rates were obtained by the

SEER program from the U.S Census Bureau The

4 years, ages 5–9 years, ages 10–14 years, and ages

15–19 years) was used in direct age standardization

Statistical Analysis

Frequencies and age-adjusted incidence rates were

incidence rates are reported as the number of cases

per 1,000,000 person-years of follow-up To examine

incidence trends the annual percent change (APC)

and 95% confidence intervals (95% CIs) were

least-squares regression, in which the independent

variable was calendar year and the dependent

vari-able was the natural logarithm of the age-adjusted

determine when the trend changed in magnitude

and/or direction during the period 1992 through

2004, allowing a maximum of 3 joinpoints and a

minimum of 2 years between consecutive joinpoints

Joinpoint permutation test adjusts for multiple

com-parisons to ensure an overall type I error rate of 0.05

Rates and trends within sex (male and female),

years, ages 10–14 years, and ages 15–19 years), race

(white, black, and Asian/Pacific Islander) and

unknown/unspecified race were excluded from sub-group analysis because there were too few cases to allow for a trend analysis

This study was approved by the University of Minnesota Institutional Review Board

RESULTS

Overall Between 1992 and 2004, 22,694 incident malignant neoplasms were reported among those patients aged

<20 years within the 13 SEER registries examined (Table 2) The average annual age-adjusted incidence rate was 158 per 1,000,000 person-years and there was a suggestion of a positive trend (APC, 0.4%; 95%

Frequencies, incidence rates, and results of the best fit joinpoint regression models for each of the cancer subtypes examined are provided in Table 2 None of the rates changed in a strictly monotonic fashion because each of the cancer diagnoses is rare and subject to random fluctuation; trends are best described by joinpoint results Rates increased sub-stantially over the time period for melanoma (APC, 2.8%; 95% CI, 0.5%–5.1%), hepatoblastoma (APC, 4.3%; 95% CI, 0.2%–8.7%), and other/unspecified central nervous system (CNS) tumors (APC, 5.4%; 95% CI, 0.8%–10.3%) Two joinpoints were found for astrocytoma; the rate decreased initially (1992–1999:

rapid increase (1999–2002: APC, 7.8%; 95% CI, 29.1%–27.9%) and a subsequent decline (2002–2004:

magnitude and direction of the trend was observed

decrease was observed between 1992 and 1996 (APC, 218.6%; 95% CI, 231.6%–23.2%) and a significant increase was detected thereafter (APC, 7.0%; 95% CI, 0.1%–14.3%) The trends for other cancer subgroups were indistinguishable from a slope of 0, although the data indicate a possible increase for leukemia overall (acute lymphoblastic leukemia [ALL] and acute myeloid leukemia [AML]) and non-Hodgkin lymphoma (NHL), and suggest a decrease for Wilms tumor and Ewing sarcoma Notably, the rate of CNS tumors overall remained steady over the time period

observed and expected rates, as determined by

join-TABLE

Incidence rate*

y (95

y (95

y (95

M 1992–2004

Incidence rate*

y (95

y (95

y (95

M 1992–2004

y (95%

y (95%

y (95%

y (95%

Female 1992–2004

y (95%

y (95%

y (95%

y (95%

Female 1992

y (95%

y (95%

y (95%

y (95%

y (95%

y (95%

Incidence rate*

y (95%

y (95%

y (95

y (95%

Incidence rate*

y (95%

y (95%

y (95

y (95%

y (95

y (95

y (95%

y (95

y (95

y (95%

y (95%

y (95%

y (95%

y (95%

y (95%

y (95%

y (95%

y (95%

In rate*

y (95

y (95

y (95

White 1992–2004

In rate*

y (95

y (95

y (95

White 1992

Incidence rate*

C1

y (95%

y (95%

y (95

y (95%

B 1992–2004

Incidence rate*

C1

y (95%

y (95%

y (95

y (95%

B 1992

In rate*

C1

y (95%

y (95%

y (95%

In rate*

C1

y (95%

y (95%

y (95%

In rate*

y (95%

C2

y (95%

y (95%

y (95%

point regression, for leukemias and CNS tumors are

shown in Figure 1

results for the demographic subgroups for each of

the cancers analyzed are listed in Table 2 The trends

were generally similar across strata of demographic

variables examined; noteworthy trends are discussed

Sex

A significantly positive trend (APC, 1.9%; 95% CI,

0.2%–3.7%) for AML was observed among males,

whereas no concomitant trend was apparent among

females For other/unspecified leukemias, the rate in

males decreased sharply between 1992 and 1997

increased between 1997 and 2004 (APC, 14.0%; 95%

CI, 2.4%–26.9%) Similarly, rapid changes punctuated

22.4%; 95% CI, 24.5%–20.3%), but not in males Joinpoints resulting in a shift in the direction of the trend also were observed among females for osteo-sarcoma (1 joinpoint), nonrhabdomyoosteo-sarcoma soft tissue sarcomas (2 joinpoints), and ependymoma (3 joinpoints); these changes were not detected among males An increase in thyroid carcinoma was sug-gested in females only

Age Group The incidence rate of ALL was stable, with the excep-tion of those ages 5 through 9 years, in whom an increase was initially observed (1992–1999: APC, 3.6%; 95% CI, 0.5%–6.8%), followed by a decline sim-ilar in magnitude, and in those patients ages 15 to

19 years, in whom a substantial increase was detected (APC, 3.3%; 95% CI, 0.4%–6.4%) Rates of HD may also

be declining among patients ages 15 to 19 years For CNS tumors overall, an initial decline was observed among those ages 1 to 4 years (1992–1996:

subse-quent rise and fall in rates; rates appeared stable in other age groups The early decline observed in the rate of ependymomas in patients ages 10 to 14 years was more rapid later in the period; this pattern was restricted to this age group and was based on few

older age groups, although rates were imprecise in all groups No change was observed in the incidence

patients ages 15 to 19 years, in whom 2 joinpoints were identified, resulting in sharp changes in the direction of the trend

A significant increase occurred in the 1990s for rhabdomyosarcoma among those ages 5 to 9 years (APC, 3.5%; 95% CI, 0.3%–6.9%), followed by a sharp decline For other soft tissue sarcomas, a significant rise was found in infants (APC, 10.4%; 95% CI, 4.9%– 16.3%), with little or no change noted in other groups An increase in patients ages 15 to 19 years was suggested for thyroid carcinoma

Race For leukemias overall and ALL, increases were observed in whites (leukemias: APC, 1.0%; 95% CI, 0.1%–2.0%; ALL: APC, 1.1%; 95% CI, 0.0%–2.3%), whereas there was no evidence of change in blacks and Asian/Pacific Islanders The trend for other leu-kemias and other CNS tumors among the white

incidence rates from Surveillance, Epidemiology, and End Results (SEER)

13 registries from 1992 through 2004 for (A) leukemias and (B) central nervous

system (CNS) tumors ALL indicates acute lymphoblastic leukemia; AML, acute

myeloid leukemia; PNET, primitive neuroectodermal tumor.

group paralleled those observed overall A decrease

and an increase in melanoma (APC, 4.2%; 95% CI,

2.2%–6.3%) were detected only among the white

group There was also a suggestion of an increase in

thyroid cancers in whites over the period examined

HD declined substantially in the black group in

subsequently rose as sharply, whereas an increase in

NHL was suggested in both whites and blacks

and osteosarcoma notably increased (APC, 4.0%; 95%

CI, 0.0%–8.1%) among the black group A statistically

significant decrease was observed in

nonrhabdomyo-sarcoma soft tissue nonrhabdomyo-sarcomas among Asian/Pacific

The trends described among blacks and Hispanics

Ethnicity

(non-Hispanic) subgroup were very similar to those

observed in the white racial group (data not shown)

An overall increase in all childhood cancers was

sug-gested among Hispanics, as well as a modest

in-crease in leukemias Unlike the fluctuating trend

observed in the non-Hispanic group, no change in

other leukemias was detected in the Hispanic group

There was a statistically significant increase in other

gliomas among Hispanic children (APC, 7.7%; 95%

CI, 1.9%–13.8%) and a substantial increase in

mela-nomas (APC, 7.5%; 95% CI, 0.4%–15.0%), which was

also observed in non-Hispanic children Rates of Ewing

sarcoma decreased between 1992 and 2000 (APC,

210.4%; 95% CI, 217.2%–23.0%) and increased

there-after (APC, 21.8%; 95% CI, 0.2%–48.1%); this trend was

DISCUSSION

that rates of childhood cancers are increasing

between 0.6% to 1.1% annually; however, these

reports include data beginning in the 1970s, and

therefore described trends may be due in large part

to rate increases in previous decades The current

analysis indicates a modest, nonsignificant increase

(0.4% annually) in all childhood cancers diagnosed

since the early 1990s, with evidence of increases in

select malignancies and shifting trends in others

The results of the current study corroborate

those from prior SEER reports, in which the overall

Although not significant, there is a suggestion of an

overall increase in leukemia, especially ALL Changes

in the rates of other/unspecified leukemias and CNS tumors, respectively, account for relatively few cases and may primarily reflect changes in classification, rather than actual decreases On further inspection

of individual ICD-O-3 codes, the decline in other leu-kemias between 1992 and 1996 is attributable to decreases in acute leukemia, not otherwise specified [NOS], and chronic myeloid leukemia, but the increase noted between 1996 and 2004 is not easily explained Fluctuations in astrocytoma were charac-terized by a decrease in astrocytoma, NOS; an increase in pilocytic astrocytoma; and an initial increase in fibrillary astrocytoma between 1992 and

2002, followed by a sharp drop The observed pattern may be attributable to random variation The increase in other/unspecified CNS tumors is not due

to the 2001 inclusion of benign tumors in the SEER

neoplasms only; a nonsignificant increase in the di-agnosis of chorioid plexus carcinomas contributed mostly to the observed increase

The rise in the rate of hepatoblastoma coincides with an increase in the frequency of low or very low

The differences by demographic subgroup are unlikely explained by changes in classification or di-agnosis because changes should have been applied equally across demographic subgroups The signifi-cant increase in AML among males, but not females,

is of interest and should be followed closely to deter-mine whether this pattern continues The decrease

in HD noted among females contradicts prior reports

of increasing rates among adolescents and young adults, but is consistent with results observed in

suggested; however, a corresponding decline in males was not detected, ruling out changes in classification

as a likely explanation The dramatic increase in ependymomas noted between 1992 and 1995 occur-ring among females involved a small number of

chance Categorizing ependymomas as borderline

is no reason to suspect classification differences by sex and across different registries A positive trend in infantile fibrosarcoma accounts for the rise in

among infants, but trend fluctuations for all CNS tumors among patients ages 1 to 4 years is not explained by specific subtypes

among white children were similar to those

described overall, although the increase in ALL and

decline in Wilms tumor are of interest If real, these

trends may reflect changes in environmental

cancer incidence among Hispanic children in

Califor-nia between 1988 and 1994 The current study results

demonstrated a significant increase in other gliomas,

which appears to be driven by an increase in

glio-mas, NOS This may be a ‘jump’ analogous to that

observed between 1984 and 1986 for other sections

improved diagnostic methods in lower

socioeco-nomic strata The observed rise in melanoma rates is

consistent with a prior report among Hispanics of all

Observed differences in trends across

demo-graphic subgroups provide starting points for

etiolo-gic research Testable hypotheses, including birth

characteristics and environmental/exogenous

expo-sures, can be investigated in relation to specific

cancers via rigorous case-control studies The

devel-opment of a national pediatric cancer registry, such

as that being implemented by the Children’s

The principal strength of SEER is the high rate of

case ascertainment and high data quality Our

analy-sis provides current trends in childhood cancer

inci-dence, and to our knowledge represents the first

report from the 13 SEER registries and the first report

regarding trends in Asian/Pacific Islanders and

His-panics One limitation of the current study is that

important demographic and/or etiologic differences

could potentially exist between the population

moni-tored by the 13 SEER registries and the 86% not

childhood cancers represent a heterogeneous group

of diseases with specific etiologies Because each

cancer diagnosis is rare, random fluctuations may

erroneously appear as noteworthy trends; a large

number of statistical tests, using an overall type I

error rate of 5% per trend, were performed in the

joinpoint regression analyses and the results

there-fore may include spurious associations Trends

trends, as well as those with wide 95% CIs, should be

interpreted cautiously Changes in classification, such

as the introduction of the International Classification

trends slightly beginning in 2001

The results of the current study indicate that inci-dence rates for pediatric cancers overall have margin-ally increased since 1992, whereas rates for select diagnoses have significantly increased Some intrigu-ing differences by sex, age, and race/ethnicity exist The trends described herein, in concert with the pau-city of information regarding underlying causes of disease, necessitate an ongoing need for population-based surveillance and further etiologic studies

REFERENCES

Atlanta, GA: American Cancer Society; 2007.

epidemiolo-gic research: lessons from studies of childhood cancer Pediatrics 2003;112(1 pt 2):218–232.

in children rises New York Times 1997;A:1.

experts/hun-dreds gather to discuss threat San Francisco Chronicle 1997;A:1.

2001;A:8.

and arsenic exposure in drinking water in Nevada Arch Environ Health 2002;57:201–206.

inci-dence and survival among children and adolescents United States SEER Program 1975–1995 NIH Pub No 99–

4649 Bethesda, Md: National Cancer Institute, SEER Pro-gram; 1999.

Can-cer surveillance series: recent trends in childhood canCan-cer incidence and mortality in the United States J Natl Cancer Inst 1999;91:1051–1058.

Can-cer Statistics Review, 1975–2004 Bethesda, Md: National Cancer Institute; 2007 Available at URL: http://seer.cancer gov/csr/1975_2004/, based on November 2006 SEER data submission, posted to the SEER website, 2007 Accessed

on June 5, 2007.

10 Surveillance, Epidemiology, and End Results (SEER) Pro-gram SEER*Stat Database: Incidence — SEER 13 Regs Limited-Use (1992–2004) Bethesda, Md: National Cancer Institute, Division of Cancer Control and Population Sciences, Surveillance Research Program, Cancer Statistics Branch, released April 2007, based on the November 2006

Ac-cessed on June 14, 2007.

11 Zippin C, Lum D, Hankey BF Completeness of hospital cancer case reporting from the SEER Program of the National Cancer Institute Cancer 1995;76:2343–2350.

12 Fritz AG, World Health Organization International Classifi-cation of Diseases for Oncology 3rd ed Geneva: World Health Organization; 2000.