Adolescent and Young Adult Oncology Program, Seattle Children's Hospital, University of Washington, Seattle, Washington
Corresponding author: Rebecca H. Johnson, MD, Seattle Children's Hospital, University of Washington, M/S MB.8.501, PO Box 5371, Seattle, WA 98145-5005; Fax: (206) 987-3946; firstname.lastname@example.org
We thank Dr. Archie Bleyer for his contributions to the development of this paper. We also thank the Seattle Children's Guild Association Teen Cancer Grant, which provided salary support for Mr. Chien.
Although rising incidence rates of testicular germ cell tumors have been well documented in white men, relatively little is known about rates in men of Hispanic origin. In the current study, the authors compared germ cell tumor trends between men of Hispanic and non-Hispanic origin as a function of age at diagnosis.
Trends in testicular germ cell tumor incidence among white men were analyzed according to Hispanic ethnicity in 2 data sets of the National Cancer Institute's Surveillance, Epidemiology, and End Results Program, spanning from 1992 to 2010 and from 2000 to 2010 and sampling 15% and 28% of the United States population, respectively. Rates were age-adjusted to the year 2000 US standard population.
Between 1992 and 2010, the annual incidence of testicular germ cell tumors among Hispanic whites ages 15 to 39 years increased 58%, from 7.18 cases per 100,000 in 1992 to 11.34 cases per 100,000 by 2010 (P<1 × 10−9). Their incidence rates increased in metropolitan areas for both seminoma and nonseminoma subtypes and for all stages at diagnosis. During the same 19-year interval, incidence among non-Hispanic white young adults increased 7%, from 12.41 to 13.22 per 100,000. During the 2000 to 2010 interval, no significant trends were observed in incidence among non-Hispanic whites.
Testicular germ cell tumors (TGCTs) are the most common solid malignancy among adolescent and young adult (AYA) men, accounting for 21% of all neoplasms in men ages 15 to 39 years according to recent data from the US Surveillance, Epidemiology, and End Results (SEER) Program. TGCTs are 1 of the most readily treatable tumors, with a 10-year survival rate of 98% for seminomas and 91% for nonseminomas. From 2005 to 2009, the overall annual incidence was 5.5 per 100,000 American men, although incidence varied widely by race. White men are at highest risk for TGCT, with an annual incidence of 6.6 per 100,000 men from 2005 to 2009, compared with only 1.4 per 100,000 in black men and 1.9 per 100,000 in Asians/Pacific Islanders during the same interval. Annual TGCT incidence in Hispanic men is 4.7 per 100,000. International data parallel incidence by race in the United States, with the highest age-adjusted incidence rates reported in Europe and Australia, where the population is predominately white.[3, 4] The incidence is lowest in Asian and African men.
Several reports in the United States have indicated increasing TGCT incidence especially among white men during the 1970s to the 1990s.[3-10] The incidence among black men began increasing in the early 1990s,[7, 10] whereas the incidence among non-Hispanic whites, Asians and Pacific Islanders, and American Indians and Alaskan Natives appeared to have plateaued at that time.[5, 10, 11]
Hispanic Americans are the fastest growing population in the United States, with an estimated 50.5 million residents according to the 2010 census. A previous report presented limited information on TGCT rates among Hispanic men in the United States from 1992 to 2003, indicating an increasing incidence that was not statistically significant. To our knowledge, there have been no subsequent reports describing trends in TGCT incidence among Hispanic Americans in the past decade. The release of the National Cancer Institute's largest and most expansive SEER data set (SEER18), along with incidence data up to 2010 in all SEER registries, allowed a more robust analysis of the trends in TGCTs among Hispanic AYAs.
MATERIALS AND METHODS
We obtained population-based data on TGCT incidence among white men of Hispanic and non-Hispanic origin from 1992 to 2010 in the SEER Program of the US National Cancer Institute.[13, 14] The original 9 SEER registries (SEER9)—consisting of Connecticut, Iowa, New Mexico, Utah, and Hawaii; the metropolitan areas of Detroit, San Francisco-Oakland, and Atlanta; and 13 counties of the Seattle-Puget Sound region—have contributed data since 1973/1974. In 1992, 4 more registries (rural Georgia, Alaskan Natives, Los Angeles, and San Jose-Monterey) were added to the SEER Program (SEER13), and Hispanic ethnicity became a required data item. For 2000 through 2008, data from 4 additional registries are available (the rest of the state of California, Kentucky, New Jersey, and Louisiana) (SEER17). In 2012, the entire state of Georgia was added retrospectively to SEER17 to constitute the SEER18 data set. Although SEER9 offers the longest period of incidence data, our analysis began with SEER13 and included SEER18, because these registries classify cases by Hispanic ethnicity, and the goal of this study was to examine TGCT incidence among men of Hispanic origin. SEER classifies individuals diagnosed with cancer according to Hispanic origin based on the North American Association of Central Cancer Registries Hispanic Identification Algorithm.
Among 42,532 white males with TGCT from the SEER13 and SEER18 data sets, 823 were aged <15 years and were excluded from this study, because disease in this age group may represent TGCT with an etiology distinct from that arising later in life. Of the remaining males, an additional 134 were excluded because their tumors were nonmalignant, and another 519 were excluded because their tumors were not first primary tumors. In total, among white males there were 18,545 TGCTs from the SEER13 data set, which included 13,187 diagnosed in the AYA age range, and 22,511 TGCTs from the SEER18 data set, which included 15,459 diagnosed in the AYA age range.
We compared TGCT incidence from 1992 to 2010 between Hispanic and non-Hispanic white men using data from both SEER13 and SEER18. Incidence trends were analyzed according to age at diagnosis in 5-year intervals. We classified AYAs as individuals diagnosed at ages 15 to 39 years, as defined by the National Cancer Institute AYA Oncology Progress Review Group. Population data are directly enumerated every 10th year by the US Bureau of the Census, and we used population estimates for the intervening years. Intercensal estimates from 1991 to 1999 and from 2001 to 2009 were updated after the release of the 2000 and 2010 national censuses, respectively. Incidence rates were age-adjusted to the US year 2000 standard population. Annual percent changes (APCs) in incidence were obtained by fitting the natural logarithm of incidence (dependent variable) to a weighted least-squares regression model, regressed over calendar year of diagnosis (independent variable). APC is a function of the estimated regression slope parameter. To assess whether any patterns observed were specific to TGCTs, we also calculated the 1992 to 2010 APCs for all malignancies and for all malignancies except TGCT among Hispanic white men ages 15 to 39 years. 95% confidence intervals were constructed from the regression slope standard error. All P values reported in this study are from tests of the null hypothesis that the corresponding APC is zero. P values were computed from a 2-tailed t distribution of the regression slope; we considered the null hypothesis to be rejected if P < .05. All APCs and P values were computed using SEER*Stat software.
We analyzed TGCT trends by histology, stage at diagnosis, and population density of the place of residence at diagnosis. We classified TGCT as either seminoma or nonseminoma using International Classification of Diseases for Oncology, third edition (ICD-O-3) morphology codes 9060 through 9062 and 9064 for seminoma and ICD-O-3 codes 9065 and 9070 through 9102 for nonseminoma. Disease stage at diagnosis was determined using the SEER Historic Stage A recode, with localized disease indicating tumors confined to the testis, regional disease indicating contiguous and adjacent organ spread, and distant disease indicating tumors that involved remote metastases. Place of residence at diagnosis was categorized as metropolitan or nonmetropolitan based on the SEER Rural-Urban continuum code 2003 developed by the US Department of Agriculture, with metropolitan counties including metropolitan areas of any population size (codes 1-3) and nonmetropolitan counties referring to urban and completely rural counties of any population size, either adjacent or not adjacent to metropolitan areas (codes 4-9).
To estimate the number of cases of TGCT among Hispanic white AYAs in the United States, the projected number of TGCT diagnoses for men in 5-year age groups was computed for each calendar year between 1992 and 2010. This number was obtained first by calculating the age-specific and year-specific ratios of the number of Hispanic white Americans in the United States to the number of AYA Hispanic white Americans in SEER13 (for 1992 through 1999) and SEER18 (for 2000 through 2010); then, this ratio was multiplied by the corresponding observed AYA SEER13 and SEER18 TGCT counts for 1992 through 1999 and 2000 through 2010, respectively. National white Hispanic population counts were retrieved from SEER.
SEER13, which spans the interval between 1992 and 2010, contains 18,545 first diagnosis, malignant, primary TGCTs in white men diagnosed at age ≥15 years, of whom 3488 were of Hispanic origin and 15,057 were of non-Hispanic origin. SEER18 (interval, 2000-2010) contains 22,511 first diagnosis, malignant, primary TGCTs in white men diagnosed at age ≥15 years, of whom 4655 were of Hispanic origin and 17,856 were of non-Hispanic origin.
Figure 1 illustrates the incidence of TGCT among white AYAs of Hispanic and non-Hispanic origin over time. TGCT incidence in Hispanic AYAs increased steadily, with an APC of 3.61 (95% confidence interval [CI], 3.02-4.21; P < × 10−9) in the interval from 1992 to 2010 for SEER13 and 3.81 (95% CI, 2.99 to 4.63; P = 2 × 10−6) in the more recent interval from 2000 to 2010 for SEER18. In contrast, among non-Hispanic AYAs, a lower APC of 1.04 (95% CI, 0.58-1.50; P = .0002) was observed in SEER13; and, in SEER18, there was no evidence of a trend in TGCT incidence. The absolute increase in TGCT incidence over the 19-year interval of SEER13 was only 0.81 cases per 100,000 non-Hispanic men per year (a proportional increase of 6.5%) compared with 4.16 cases per 100,000 Hispanic men per year (a proportional increase of 57.9%).
Figure 2 plots SEER13 APCs for testicular cancer incidence by 5-year age intervals among white men of Hispanic and non-Hispanic origin ages 15 to 49 years. (APCs were not calculated for Hispanic men aged >49 years because of insufficient sample size.) Among Hispanics, we observed statistically significant increases in APC exceeding 1.5 for all 5-year age groups in the AYA age range (ages 15-39 years). No APC values significantly greater than zero were observed in Hispanic men aged >39 years. In non-Hispanic men, we observed a statistically significant but much smaller APC increase in the subset of AYAs between ages 20 and 34 years, but not in other age groups. APCs that reached statistical significance (P < .05) among non-Hispanic white AYA men ranged from 0.9 to 1.6.
Table 1 compares TGCT incidence trends for AYAs and older men of Hispanic and non-Hispanic origin according to histology, stage at diagnosis, and metropolitan versus nonmetropolitan residency in SEER13 and SEER18. For Hispanic white men, TGCT incidence in AYAs increased for both seminoma and nonseminoma subtypes and for all stages at diagnosis. The APC for seminoma and nonseminoma among Hispanic AYAs in SEER13 were 3.07 (95% CI, 1.94-4.21; P = .00,002) and 4.10 (95% CI, 2.91-5.32; P = 1 × 10−6), respectively. The incidence of localized disease increased at an APC of 3.50 (P = 5 × 10−7), the incidence of regional disease increased at an APC of 4.49 (P = 0.003), and the incidence of distant disease increased at an APC of 2.26 (P = .02). The increasing trend in TGCT incidence occurred only in metropolitan counties, where the APC was 3.70 (95% CI, 3.05-4.34; P = 1 × 10−9) in SEER13. In SEER18, significant and positive APCs were demonstrated among Hispanic AYAs in every category of histology and stage at diagnosis analyzed and in both metropolitan and nonmetropolitan counties, with a relatively high APC of 6.78 (95% CI, 0.21-13.78; P = .04) observed in nonmetropolitan areas. In contrast to Hispanic AYAs, the corresponding APCs among older Hispanic white men ranged from strongly negative to strongly positive.
Table 1. Average Annual Percent Change in Testicular Germ Cell Tumor Incidence Among Whites, by Hispanic Ethnicity and Age, in the Surveillance, Epidemiology, and End Results (SEER) 13 Registries (SEER13) and SEER18 Databases
AYAs (Aged 15-39 Years)
Older Men (Aged 40+ Years)
APC (95% CI)
APC (95% CI)
Abbreviations: APC, annual percent change; AYAs, adolescents and young adults; CI, confidence interval; No., number of patients with testicular germ cell tumors.
The APC could not be calculated because of insufficient sample size.
Within the AYA age group, non-Hispanic whites had a consistently lower APC than in Hispanic men. In both SEER data sets, the APC in non-Hispanic whites did not exceed 1.5. In SEER18, no APCs from non-Hispanic white AYAs reached statistical significance, except for distant disease at diagnosis, for which an APC of 3.15 was observed in SEER18 (95% CI, 0.93-5.43; P = .01).
The overall incidence of cancer among AYA Hispanic Americans between 1992 and 2010 did not show a statistically significant increase or decrease in SEER13 (data not shown). In total, 10,423 cancers were not of TGCT origin, and 2993 were TGCTs. Non-TGCTs demonstrated a decreasing trend in incidence (APC, −1.62; 95% CI, −2.55 to −0.69; P = .002). We estimate that, in 2010, 1134 Hispanic men in the United States were diagnosed with TGCT compared with 410 in 1992.
The incidence of TGCT in the United States increased among Hispanic white AYAs in the interval between 1992 and 2010, in contrast to a plateau in the incidence for non-Hispanic white men during the same period. In SEER13, the incidence of TGCT among Hispanic AYAs rose 1.6-fold, from 7.2 per 100,000 Hispanic men in 1992 to 11.3 per 100,000 in 2010. In SEER18, with the larger sample size and representing only the most recent years of diagnosis, we observed a similar rise in TGCTs among Hispanic AYAs (1.4-fold) over the course of 11 years. The increase affected every 5-year age group between ages 15 and 39 years, for both seminoma and nonseminoma subtypes, and for every stage at diagnosis. Comparable increases were not observed among non-Hispanic white AYAs or among older men, although the relatively smaller number of older men with TGCT indicates that the power to detect trends in this group was limited.
The increasing incidence of TGCT among Hispanics potentially may be caused by improvements in the classification of Hispanics in SEER data over time, because the North American Association of Central Cancer Registries Hispanic Identification Algorithm relies on input data (eg place of birth, surname), which may have become more complete or accurate in recent years. For such phenomena to explain our results, the improvements in classification of Hispanics (and, thus, increases in incidence) would be expected to affect all cancers as opposed to being specific to TGCT. Yet the incidence rate of all cancers excluding TGCTs among Hispanic AYAs, a sample size that outnumbers TGCTs by a factor of 3.5, appears to have decreased, rather than increased, over time.
The Hispanic population is the fastest growing ethnic minority in the United States, accounting for more than half of the total population growth in the United States between 2000 and 2010.12 This rapid rate of growth has posed challenges to the accuracy of the decennial enumeration of Hispanics and to intercensal estimates of the Hispanic population, which, in turn, may have impacted our analysis of SEER TGCT incidence trends and projections of TGCT cases among US Hispanics. Census Bureau studies indicate that, in the 1990, 2000, and 2010 decennial censuses, Hispanics were undercounted by 5%, 0.7%, and 1.5%, respectively.[19, 20] Because intercensal estimates are based heavily on the decennial censuses, during our study period from 1992 to 2010, Hispanic denominators are smaller than they should be, and more so in the earlier years than more recent years. This pattern would lead to observed estimated APCs that are weaker than the true APCs and a projected number of US Hispanics with TGCT that is smaller than it would be had Hispanics been accurately counted. Thus, the undercounting of the Hispanic population cannot explain our findings and, if anything, suggests that the true increase in TGCT incidence is greater than we have reported.
Hispanics are a heterogeneous population: the majority of Hispanic Americans in 2010 were of Mexican descent (63%) followed by those of Puerto Rican descent (9.2%). The remaining Hispanic Americans are represented by numerous ethnic subgroups from Central and South America. Future research should assess whether the rising TGCT trends are similar across subgroups of Hispanic ethnicities.
The reasons for the increasing TGCT incidence among Hispanic white AYAs, like those that underlie the increasing TGCT rates previously observed in other racial and ethnic groups in the United States, are currently unknown. The 4 established risk factors for TGCT are height,[21-26] cryptorchidism,[27, 28] family history of germ cell tumors, and prior TGCT. Men with undescended testis are at a 2.5-11 fold increased risk for TGCT compared with the general population. Although there is some evidence that the incidence of cryptorchidism in the United States increased from the late 1960s through the middle 1990s, we are not aware of any reports on the trends in this predisposing condition within specific racial or ethnic groups.
Changes in adult height are strongly affected by early life nutrition. Recent increases in TGCT incidence in different race and ethnicity groups over different calendar periods may be reflective of racial-specific and ethnic-specific secular changes in improved access to adequate nutrition. Particularly relevant to our findings, there is evidence that over the 30-year period during which most patients with TGCT in the current study were children, the height of Mexican American boys increased to a much greater extent than for white boys or black boys.
Researchers have recently reported a positive association between marijuana use and TGCT risk, in particular for nonseminomas.[31-33] Hispanics are currently more likely than whites or African Americans to use marijuana. In 2011, 50% of Hispanic teens reported marijuana use compared with 40% of African American teens and 35% of Caucasian teens. Since the early 1990s, the prevalence of lifetime marijuana use among Hispanic 12th graders has increased, from 36.6% in 1994 to 48.7% in 2012.35 Increases also were noted in non-Hispanic white and black 12th graders; although, since 2008, there has been a disproportionate acceleration in marijuana use among Hispanics.[34, 35] Yet because marijuana use has increased among both Hispanic and non-Hispanic whites, and because the incidence of both seminoma and nonseminoma TGCT rose among Hispanic whites, marijuana use alone does not adequately explain the observed increase in TGCT among Hispanic AYAs.
In addition to the risk factors for TGCT noted above, several types of environmental exposures have been studied for a potential role in TGCT pathogenesis. These include heat, polyvinylchloride, nonionizing radiation, heavy metals, agricultural work, pesticides, and polychlorinated biphenyls. For none of these, however, is there evidence clearly supporting an association with TGCT. Similarly, current evidence does not support links between TGCT risk and diet, physical activity, or exposure to ionizing radiation. Obesity has increased in the United States over the past several decades,[36-38] and trends among Hispanics have mirrored those of other racial groups. However, a high body mass index is associated, if anything, with a reduced risk of TGCT, not an increased risk.[40, 41]
In summary, we compared TGCT incidence trends in Hispanic and non-Hispanic white men in the United States. From 1992 to 2010, the incidence of these malignancies has increased among AYA Hispanic whites; whereas, among AYA non-Hispanic whites, the incidence has plateaued. The observed rising incidence of TGCTs, in combination with the increase in the Hispanic American population, predicts an increasing clinical importance of TGCTs in the United States. Although, as of 2010, AYA TGCT incidence has remained lower in Hispanic whites than in non-Hispanic whites, the rate of TGCT among Hispanic whites may overtake that among non-Hispanic whites if the observed trends persist. Future studies should corroborate these results and investigate the factors responsible for the increase.
Mr. Chien's salary was funded by a Seattle Children's Guild Association Teen Cancer Grant. Salaries for Dr. Schwartz and Dr. Johnson were provided by their respective institutions.