Racial/Ethnic Differences in Cancers Attributable to Preventable Infectious Agents in Texas, 2015

Abstract

Objective

The International Agency for Research on Cancer has classified 13 infectious agents as carcinogenic or probably carcinogenic to humans. We aimed to estimate the percentage (ie, population-attributable fraction) and number of incident cancer cases in Texas in 2015 that were attributable to oncogenic infections, overall and by race/ethnicity.

Methods

We calculated population-attributable fractions for cancers attributable to human papillomavirus (HPV), Helicobacter pylori, hepatitis C virus (HCV), hepatitis B virus (HBV), and human herpesvirus 8 (HHV-8) infections using prevalence estimates from National Health and Nutrition Examination Survey laboratory data and relative risks associated with infection from previous epidemiological studies. The Texas Cancer Registry provided cancer incidence data.

Results

We estimated that 3603 excess cancer cases, or 3.5% of all cancers diagnosed in 2015, among adults aged ≥25 in Texas were attributable to oncogenic infections. Hispanic adults had the highest proportion of cancer cases attributable to infections (5.6%), followed by non-Hispanic Black (5.4%) and non-Hispanic White (2.3%) adults. HPV infection caused the highest proportion of all cancer cases (1.8%) compared with other oncogenic infections (HCV, 0.8%; H pylori, 0.5%; HBV, 0.3%; HHV-8, 0.1%). Hispanic adults had the highest proportions of all cancers caused by HPV infection (2.6%) and H pylori (1.1%), and non-Hispanic Black adults had the highest proportions of all cancers caused by HCV infection (1.7%), HBV infection (0.7%), and HHV-8 (0.3%).

Conclusion

Preventable oncogenic infections contribute to cancer incidence in Texas and may affect racial/ethnic minority groups disproportionately. Infection control and prevention should be stressed as an important component of cancer prevention.

The International Agency for Research on Cancer (IARC) classified 13 infectious agents as carcinogenic (Group 1) or probably carcinogenic (Group 2A) to humans in its Monographs on the Evaluation of Carcinogenic Risks to Humans: Epstein-Barr virus, hepatitis B virus (HBV; chronic infection), hepatitis C virus (HCV; chronic infection), human herpesvirus 8 (HHV-8), HIV-1, human papillomaviruses (HPV; types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68), human T-cell lymphotropic virus type 1 (HTLV-1), Opisthorchis viverrini (chronic infection), Clonorchis sinensis (chronic infection), Schistosoma haematobium (chronic infection), Helicobacter pylori (chronic infection), Plasmodium falciparum, and Merkel cell polyomavirus.^1,2 In the United States, O viverrini, C sinensis, S haematobium, P falciparum, and HTLV-1 are not endemic,^1,2 but the other infectious agents pose cancer risks. Furthermore, in the United States, the incidence rates of cancers associated with certain infectious agents, such as HPV-associated cancers, differ by race/ethnicity.³ Because these oncogenic infections are potentially preventable, recognizing racial/ethnic disparities in infection-attributable cancers is critical in establishing effective cancer prevention programs in the United States and its member states.

The objective of our analysis was to estimate the population-attributable fractions (PAFs) and number of cancer cases diagnosed in Texas in 2015 that were attributable to relevant endemic oncogenic infections. We defined PAF as the percentage of new cancers diagnosed in Texas in 2015 that could theoretically have been avoided if these oncogenic infections were prevented. In addition, we stratified our PAF analysis by race/ethnicity to reveal any disparities that may exist in the proportions of cancers caused by oncogenic infections across major racial/ethnic subgroups in Texas.

Methods

We analyzed cancers caused by HPV (strain 16 only: cancers of the vulva, vagina, penis, anus, oral cavity, oropharynx/tonsil; all oncogenic strains: cervical cancer), chronic H pylori (noncardia stomach cancer, gastric mucosa-associated lymphoid tissue lymphoma [MALToma]), chronic HCV (hepatocellular carcinoma [HCC], non-Hodgkin lymphoma), chronic HBV (HCC), and HHV-8 (Kaposi sarcoma; we excluded primary effusion lymphoma because of lack of representative incidence data). The oncogenic potential of HIV-1 is largely attributable to coinfection with other oncogenic infectious agents as a result of immunosuppression, so we did not analyze HIV-1 separately, in accordance with previous studies.⁴ We omitted Epstein-Barr virus and Merkel cell polyomavirus because of lack of representative prevalence data.

The Texas Cancer Registry provided data on the numbers of invasive cancers diagnosed in Texas in 2015, overall and by age group (25-34, 35-44, 45-54, 55-64, 65-74, 75-84, ≥85), sex (male, female), and race/ethnicity (all racial/ethnic groups, non-Hispanic White, non-Hispanic Black, Hispanic, other racial/ethnic groups).⁵ For HPV, we set the 2 oldest age categories at 65-69 and ≥70 because of limitations in prevalence data. We used the Surveillance, Epidemiology, and End Results (SEER) Site Recode International Classification of Diseases (ICD)-O-3/World Health Organization 2008 definition to code cancer sites.⁶ We coded noncardia stomach cancer⁷ and gastric MALToma⁶ with primary ICD-O-3 site codes, and we further specified gastric MALToma and HCC with histology codes (Table 1).¹¹ Because the data were de-identified and aggregated and are publicly available, the institutional review boards at the Texas Cancer Registry and Baylor College of Medicine considered this study exempt.

Table 1.

Relative risks of cancer development associated with seropositive status, by infectious agent and cancer site, for all adults and by sex^a

Infectious agent and cancer site	Relative risk
	Men	Women	All adults
Human papillomavirus strain 16 (HPV-16)
Vulva	—	3.70	—
Vagina	—	6.30	—
Penis	11.60	—	—
Anus	5.30	5.90	—
Oral cavity	—	—	1.94
Oropharynx, tonsil	—	—	8.60
Cervixb	NAd	NAd	NAd
Helicobacter pylori
Stomach, noncardia (primary site codes: C161-166)	—	—	5.90
Gastric MALTomac (primary site codes: C160-169; histology code: 9699)	—	—	7.20
Hepatitis C virus
Hepatocellular carcinoma (histology codes: 8170-8175)	—	—	27.60
Non-Hodgkin lymphoma	—	—	1.78
Hepatitis B virus
Hepatocellular carcinoma (histology codes: 8170-8175)	—	—	23.40
Human herpesvirus type 8
Kaposi sarcoma	NAd	NAd	NAd

Abbreviations: —, denotes that infectious agent is not causally associated with cancer site, and no relative risk was collected; MALToma, mucosa-associated lymphoid tissue lymphoma; NA, not applicable.

^aData sources: Plummer et al,⁴ de Martel et al,⁸ Carter et al,⁹ and Anantharaman et al.¹⁰

^bCervical cancer is associated with HPV-16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68, as opposed to the other cancer sites listed for HPV-16, which are associated only with HPV-16 infection.

^cRelative risk provided is for combined gastric MALToma and gastric diffuse large B cell lymphoma.

^dRelative risk is not applicable because the population-attributable fraction is assumed to be 100%.

We collected data on estimates of relative risk (RR) associated with seropositivity for infection from similar PAF studies, which used pooled analyses of cohort or case-control studies as available (Table 1)^4,8; otherwise, we used odds ratios from case-control studies (pooled or unpooled) as estimates for RRs.^9,10 We used RR as a measure of the risk of cancer development among adults with infection compared with adults without infection. Consistent with other studies, we assumed the PAF to be 100% for HPV-associated cervical cancer and HHV-8–associated Kaposi sarcoma; as such, we did not collect RRs.⁴

We sourced weighted prevalence estimates from National Health and Nutrition Examination Survey (NHANES) laboratory data,^12,13 which represent the US Census civilian noninstitutionalized population, weighted to account for oversampling, nonresponse, and post-stratification.¹⁴ Given that no Texas-specific database was available, we assumed that US-wide data were representative of Texas. We assumed a latency period of 10 years from infection to cancer diagnosis for HBV, HCV, and HPV, a conservative estimate based on previous reports.^15,16 Thus, we used NHANES data from survey years 2005-2006 for these infectious agents. NHANES defines chronic HBV infection as seropositivity for hepatitis B surface antigen (HBsAg) for 6 months; we used positive HBsAg as a marker for infection.¹⁷ NHANES defines chronic HCV infection as seropositivity for anti-HCV antibodies and detectable HCV RNA for 6 months.¹⁸ In this analysis, we considered HCV infection positive if anti-HCV was either positive or indeterminate and HCV RNA was positive.¹⁹ We defined HPV infection by positive results from the 9-plex Competitive Luminex Assay using L1-viral-like particles.¹² We collected prevalence data only for HPV-16, which is associated with 7 cancer sites, including cervical cancer.¹ Because the 12 other oncogenic HPV strains identified by IARC as being causally related to cancer are associated only with cervical cancer,¹ for which the PAF is assumed to be 100%, we did not collect prevalence data for these strains because we did not calculate PAFs. However, for HPV-16, we had prevalence data only until age 60; thus, we estimated prevalence data for adults aged ≥60 from the prevalence for adults aged 50-59. For H pylori, the latency period is unknown but prolonged; infection usually begins in childhood or earlier.²⁰ For consistency, we also assumed a 10-year latency period for H pylori. However, we had prevalence data only for NHANES survey years 1999-2000. Given that the prevalence of H pylori infection in the United States did not change substantially from 1990 to 2006,²¹ we applied the 1999-2000 prevalence data as an estimate for 2006 prevalence rates. We used seropositivity for anti–H pylori immunoglobulin to determine prevalence.¹³ For HHV-8, we did not collect prevalence data because the PAF was assumed to be 100%. We stratified prevalence data by age group, sex, and race/ethnicity.

We used the standard formula to calculate PAFs for each infection and each cancer type by age group, sex, and race/ethnicity: $PAF=\frac{\sum \left(p*ERR\right)}{1+\sum \left(p*ERR\right)}$ , where p is the proportion of the population with infection and ERR is excess RR (ie, ERR = RR – 1).²² Here, the ERR is the extra risk of cancer development among adults with infection as compared with adults without infection when background risk factors are removed. The PAF is the estimated percentage of new cancer cases in the specified population that are due to the risk factor. We multiplied incident cancer counts by derived PAFs to calculate the number of excess cases attributable to each infection, stratified by age group, sex, and race/ethnicity. Excess cases are the excess number of cases in the population that could have been avoided if the risk factor were eliminated. To account for 10-year latency, we paired prevalence data and associated PAFs in each age group with the cancer incidence age group 10 years older (eg, 2006 prevalence data for age group 25-34 years were paired with 2015 cancer counts for age group 35-44 years). We calculated age-weighted percentages of incident cancer cases attributable to infection by cancer type, sex, and race/ethnicity. We calculated the percentages of HCC cases attributable to HBV and HCV as a percentage of all incident liver cancers, in accordance with other studies.²² Similarly, we calculated noncardia stomach cancers and gastric MALToma cases attributable to H pylori as percentages of all incident stomach cancers or non-Hodgkin lymphoma cancers, respectively.²³ For each infection, we summed all cancer types to calculate the percentage of all incident cancers (excluding basal cell carcinoma [BCC] and squamous cell carcinoma [SCC] of the skin) that were attributable to that infection. Lastly, we summed all cancers attributable to all 5 infectious agents to calculate the percentage of all incident cancers diagnosed in Texas adults aged ≥25 in 2015 attributable to oncogenic infections.

Results

Among Texas adults aged ≥18 in 2006, the estimated overall seroprevalence of HPV-16 infection was higher than the prevalence of H pylori, HCV, and HBV infections (Table 2). Men had a higher prevalence of H pylori, HCV, and HBV infections than women, and women had more HPV-16 infections than men. By race/ethnicity, non-Hispanic Black adults had the highest prevalence of HPV-16, HCV, and HBV infections, and Hispanic adults had the highest prevalence of H pylori infections.

Table 2.

Prevalence of oncogenic infections among adults aged ≥18, by race/ethnicity, National Health and Nutrition Examination Survey (NHANES) laboratory data, United States, 2006

Race/ethnicity	Men				Women				All adults
	HPV-16 (L1)a	H pylori (IgG)b	HCV (RNA + anti-HCV)c	HBV (HBsAg)d	HPV-16 (L1)a	H pylori (IgG)b	HCV (RNA + anti-HCV)c	HBV (HBsAg)d	HPV-16 (L1)a	H pylori (IgG)b	HCV (RNA + anti-HCV)c	HBV (HBsAg)d
All racial/ethnic groups	5.2	30.6	1.6	0.5	15.3	29.8	0.6	0.2	10.3	30.1	1.1	0.4
Non-Hispanic White	5.4	19.7	1.6	0.4	15.3	21.5	0.2	0.1	10.4	20.6	0.9	0.2
Non-Hispanic Black	6.2	55.9	2.6	1.1	23.3	48.4	2.5	0.9	15.4	51.9	2.5	1.0
Hispanic	3.6	62.6	1.0	0.2	10.1	53.0	0.2	0.0	6.8	57.5	0.6	0.1
Other racial/ethnic groups	5.2	45.8	0.0	3.1	8.9	38.0	2.1	0.1	7.3	41.9	1.2	1.4

Abbreviations: anti-HCV, antibody to hepatitis C virus; H pylori, Helicobacter pylori; HBsAg, hepatitis B virus surface antigen; HBV, hepatitis B virus; HCV, hepatitis C virus; HPV-16, human papillomavirus strain 16; IgG, immunoglobulin G; RNA, ribonucleic acid.

^aHPV-16 infection is defined by seropositivity for L1-viral-like-particles.

^b H pylori infection is defined by seropositivity for H pylori IgG antibodies. Data source: NHANES 1999-2000.¹³

^cChronic HCV infection is defined by seropositivity for anti-HCV antibodies and detectable HCV RNA for 6 months.

^dChronic HBV infection is defined by seropositivity for HBsAg for 6 months.

In 2015, 51 472 men and 51 936 women aged ≥25 in Texas were diagnosed with cancer (excluding BCC and SCC of the skin). Most cancer cases were among non-Hispanic White adults (n = 65 214 cases), followed by Hispanic adults (n = 22 642 cases), non-Hispanic Black adults (n = 12 020 cases), and adults in other racial/ethnic groups (n = 3532 cases). Overall, 3.5% of all incident cancer cases, or 3603 excess cases (excluding BCC and SCC of the skin), were attributable to oncogenic infections in 2015 (Table 3). Thus, 3.5% of cancers (representing >3600 cancer patients) in Texas could have been prevented if oncogenic infections had been prevented. Women had a higher proportion of all cancers attributable to infections than men (3.7% vs 3.3%). The infection with the highest PAF for all cancers combined was HPV (1.8%), followed by chronic HCV (0.8%), chronic H pylori (0.5%), chronic HBV (0.3%), and HHV-8 (0.1%); in other words, HPV infection, compared with other infectious agents, contributed the greatest proportion of preventable cancers. Overall infection-specific PAFs for HPV were numerically higher among women than men, whereas PAFs for other infections were numerically higher among men than women. Aside from cervical cancer and Kaposi sarcoma, the cancer sites with the highest site-specific PAFs for each infection were stomach for chronic H pylori, liver for chronic HCV infection, and liver for chronic HBV infection (only cancer site). For all infections combined, the sites with the highest calculated PAFs (excluding assumed PAFs for cervical cancer and Kaposi sarcoma) were liver, vagina, and oropharynx/tonsil (Table 4).

Table 3.

Age-weighted population-attributable fractions (PAFs, shown as percentages) of cancers attributable to oncogenic infections in Texas in 2015 among adults aged ≥25, by race/ethnicity, infection, and cancer site^{a , b}

Characteristic	Human papillomavirusc								Chronic Helicobacter pylori			Chronic hepatitis C virus			Chronic hepatitis B virus		Human herpesvirus type 8		All infections
	Vulva	Vagina	Penis	Anus	Oral cavity	Oropharynx, tonsil	Cervix	All cancers,d PAF (no. of excess cases)	Stomach	Non-Hodgkin lymphoma	All cancers,d PAF (no. of excess cases)	Liver	Non-Hodgkin lymphoma	All cancers,d PAF (no. of excess cases)	Liver	All cancers,d PAF (no. of excess cases)	Kaposi sarcoma	All cancers,d PAF (no. of excess cases)	All cancers,d PAF (no. of excess cases)
All
Men	—	—	31.0	17.3	4.2	28.1	—	0.6 (296)	25.6	1.1	0.6 (287)	32.5	1.2	1.4 (734)	14.0	0.6 (303)	100.0	0.2 (85)	3.3 (1705)
Women	22.7	34.7	—	34.7	9.9	47.2	100.0	2.9 (1523)	31.4	1.1	0.5 (246)	10.9	0.4	0.2 (92)	3.2	0 (24)	100.0	0 (14)	3.7 (1899)
All adults	22.7	34.7	31.0	27.3	5.9	31.3	100.0	1.8 (1818)	28.0	1.1	0.5 (533)	26.9	0.9	0.8 (826)	11.1	0.3 (327)	100.0	0.1 (99)	3.5 (3603)
Non-Hispanic White
Men	—	—	29.0	16.6	3.9	27.7	—	0.6 (213)	15.3	0.8	0.3 (86)	27.4	0.9	0.8 (274)	10.5	0.3 (99)	100.0	0.1 (20)	2.1 (692)
Women	21.6	32.1	—	33.3	9.4	45.4	100.0	2.2 (711)	25.2	1.1	0.3 (86)	1.6	0.1	0 (6)	0.1	0	100.0	0 (4)	2.5 (807)
All adults	21.6	32.1	29.0	26.7	5.4	30.6	100.0	1.4 (923)	19.0	0.9	0.3 (172)	20.7	0.6	0.4 (280)	7.8	0.2 (100)	100.0	0 (24)	2.3 (1499)
Non-Hispanic Black
Men	—	—	44.1	22.4	6.3	33.5	—	0.5 (31)	38.7	2.0	0.9 (56)	52.1	2.8	2.8 (167)	24.2	1.3 (75)	100.0	0.4 (23)	5.9 (352)
Women	38.1	47.9	—	50.4	16.2	62.4	100.0	3.2 (191)	42.4	0.9	0.8 (49)	31.1	2.1	0.6 (35)	11.9	0.2 (12)	100.0	0.1 (8)	4.9 (295)
All adults	38.1	47.9	44.1	31.8	10.2	39.8	100.0	1.9 (223)	40.4	1.5	0.9 (105)	47.0	2.5	1.7 (202)	21.2	0.7 (87)	100.0	0.3 (31)	5.4 (648)
Hispanic
Men	—	—	21.9	11.3	2.8	19.0	—	0.3 (32)	38.7	1.6	1.4 (145)	35.0	1.7	2.7 (293)	7.4	0.6 (60)	100.0	0.3 (37)	5.3 (567)
Women	17.5	32.6	—	29.9	7.1	40.1	100.0	4.7 (560)	38.8	1.2	1.0 (114)	4.7	0.2	0.1 (15)	0	0	100.0	0 (2)	5.8 (691)
All adults	17.5	32.6	21.9	22.6	4.3	21.9	100.0	2.6 (592)	38.8	1.4	1.1 (259)	26.6	1.0	1.4 (308)	5.3	0.3 (60)	100.0	0.2 (39)	5.6 (1258)
Other racial/ethnic groups
Men	—	—	0	36.6	3.5	19.9	—	0.3 (6)	43.3	1.9	1.4 (23)	0	0	0	24.8	1.5 (25)	100.0	0.3 (5)	3.5 (59)
Women	8.1	27.0	—	0	6.2	32.8	100.0	2.8 (51)	41.9	0	0.8 (15)	15.1	2.4	0.3 (6)	0.7	0	100.0	0	3.9 (72)
All adults	8.1	27.0	0	36.6	4.5	23.7	100.0	1.6 (57)	42.7	1.1	1.1 (38)	3.6	1.0	0.2 (6)	19.0	0.7 (25)	100.0	0.1 (5)	3.7 (131)

Abbreviation: —, denotes that infectious agent is not causally associated with cancer site, and no PAF was calculated.

^aAll cancers combined are displayed as: PAF (number of excess cases).

^bTotals may not sum because of rounding.

^cCancers of the vulva, vagina, penis, anus, oral cavity, and oropharynx/tonsil are associated only with human papillomavirus strain 16 (HPV-16); cervical cancer is associated with HPV strains 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68.

^dExcluding basal cell carcinoma and squamous cell carcinoma of the skin.

Table 4.

Cancer sites ranked from highest to lowest population-attributable fraction (PAF) for all analyzed oncogenic infections combined,^a for all people aged ≥18 and all racial/ethnic groups combined, Texas, 2015

Cancer site	PAF, %
Cervixb	100.0
Kaposi sarcomab	100.0
Liver	38.0
Vagina	34.7
Oropharynx, tonsil	31.3
Penis	31.0
Stomach	28.0
Anus	27.3
Vulva	22.7
Oral cavity	5.9
Non-Hodgkin lymphoma	2.0

^aHuman papillomavirus, chronic Helicobacter pylori, chronic hepatitis C virus, chronic hepatitis B virus, and human herpesvirus type 8.

^bPAFs for cervical cancer and Kaposi sarcoma were assumed to be 100%.

When stratified by race/ethnicity, 5.6% of all cancers among Hispanic adults and 5.4% among non-Hispanic Black adults were attributable to infections, compared with only 2.3% among non-Hispanic White adults (Table 3). Overall, women had a numerically higher PAF for all cancer cases combined than men for all racial/ethnic groups, except non-Hispanic Black adults. Compared with adults in other racial/ethnic groups, Hispanic adults had numerically higher percentages of cancers attributable to HPV and chronic H pylori infections. Non-Hispanic Black adults had numerically higher infection-specific PAFs for chronic HCV, chronic HBV, and HHV-8 infections. Across all racial/ethnic groups, the overall PAF for all cancers attributable to HPV infection was numerically higher among women than men; for the rest of infections, overall infection-specific PAFs were numerically higher among men than women across all racial/ethnic groups, except for chronic H pylori among non-Hispanic White (men and women equal) and chronic HCV infection among adults in other racial/ethnic groups.

Discussion

Our findings are similar to findings from other studies. In the United States, an estimated 3.3% of all incident cancers were attributable to infections among adults aged ≥30 in 2014¹⁹; in Australia, 2.9% of all cancers were attributable to infections in 2010²²; and in the United Kingdom, this estimate was 3.6% in 2015.²³ When stratified by sex, we found that women (3.7%) had a numerically higher proportion of cancers attributable to oncogenic infections than men (3.3%). This finding is consistent with studies in Australia (women, 3.7%; men, 2.4%)²² and the United Kingdom (women, 4.2%; men, 3.1%),²³ and it likely reflects the disproportionately higher incidence of HPV-attributable cancers among women than among men, which represent most infection-attributable cases. However, a direct comparison of overall PAFs across studies is challenging, because each study used a different methodology. Most notably, these studies included additional infections that were excluded from the current analysis for reasons described previously; the US study included HIV,¹⁹ whereas the Australian²² and UK²³ studies included Epstein-Barr virus and HIV. Furthermore, our analysis included slightly different cancer sites than the other studies because of the inclusion of different infections and limitations in the available data sources.

When we analyzed data by infectious agent, we found that HPV contributed the highest overall percentage of cancer cases (1.8%) compared with other agents. This finding is similar to the findings of studies conducted in the United States (1.8%),¹⁹ Australia (1.5%),²² and the United Kingdom (1.6%)²⁴ in 2010. The finding is likely related to HPV having oncogenic potential at more cancer sites than other infectious agents, and most HPV-attributable cases occur in the cervix, which has an assumed PAF of 100% and substantially influences the overall infection-specific PAF.

To the best of our knowledge, no other studies have examined PAFs for oncogenic infections separately for the major racial/ethnic subgroups of Texas. We estimated that a numerically higher proportion of excess cancer cases in Texas were attributable to oncogenic infections among Hispanic (5.6%) and non-Hispanic Black (5.4%) people compared with non-Hispanic White people (2.3%). Furthermore, Hispanic people had the highest proportions of all cancers attributable to HPV and H pylori, whereas non-Hispanic Black people had the highest proportions of all cancers attributable to HCV, HBV, and HHV-8. The higher prevalence of HCV and HBV infections among non-Hispanic Black (HCV, 2.5%; HBV, 1.0%) than among Hispanic (HCV, 0.6%; HBV, 0.1%) people likely explains why non-Hispanic Black people had numerically more cancers attributable to these infections than Hispanic people. Similarly, the higher prevalence of H pylori among Hispanic (57.5%) than among non-Hispanic Black (51.9%) people explains the numerically higher overall PAF for H pylori among Hispanic than non-Hispanic Black people. However, our reported prevalence of HPV-16 among non-Hispanic Black people (15.4%) was notably higher than among Hispanic people (6.8%), whereas the overall PAF for HPV was higher among Hispanic (2.6%) than among non-Hispanic Black (1.9%) people. We cannot speculate on the reason for this discrepancy because we did not calculate the prevalence of HPV infection in these populations (including all other strains associated with cervical cancer) in our analysis. The finding that Hispanic people had a numerically higher overall PAF than non-Hispanic Black people for all cancers combined is likely because Hispanic people had the highest PAF for HPV, which contributes about half of all infection-attributable cancer cases, and the highest PAF for H pylori, which is the most prevalent infectious agent in the United States.

Limitations

Our study had several limitations. First, because this was a descriptive study, any differences reported between subgroups are purely numerical rather than significant; thus, findings should be interpreted with caution. Second, given that estimates were derived from a single year and single source, and that small numbers of cancers were reported in racial/ethnic minority populations, our estimates may be less precise than if the number of cases were larger; real differences between racial/ethnic subgroups may differ from those observed in our study, which may be the result of chance. Third, because of lack of endemicity or available data sources, we omitted from analysis various infectious agents known to be oncogenic. However, given the large and diverse population of immigrants in Texas, it is possible that we excluded a few rare cases attributable to nonendemic infections in Texas, resulting in a slight underestimation of the overall percentage of cancers attributable to all oncogenic infections. Fourth, given that a Texas-specific database for infection prevalence was unavailable, we assumed that US-wide prevalence data from NHANES were representative of the Texas population; however, differences in seroprevalence estimates may exist between the 2 populations.

Fifth, as with other studies,²² we also assumed a latency period of 10 years from exposure to outcome, although latency periods have not been clearly elucidated in the literature recently and likely differ by infectious agent and cancer site. Sixth, we applied standard formulae to calculate PAFs using seroprevalence for all infectious agents because of data availability from NHANES for the study years examined. However, for HPV, previous studies used the prevalence of HPV DNA in tumor cells to assess HPV-attributable cancers, because viral presence suggests an infectious cause of cancer.^19,22,24 It is possible that the application of standard formulae using seroprevalence for HPV is not the best methodology, but DNA-based prevalence data were unavailable for both sexes for 2005-2006. In addition, IARC’s conclusions concern chronic infection for several infectious agents; because of available data, we used seropositivity for prevalence estimates, which does not differentiate between chronic and acute infection, or current and previous infection, especially because seropositivity was measured only at 1 time point. Thus, seropositivity may not be the most accurate representation of infection prevalence. Lastly, we obtained estimates of RR, which form the backbone of the current study, from available literature. Inherent bias in observational studies, including measurement error and bias by unobserved individual characteristics, may have influenced our PAFs. The overall direction of the bias is unclear given the multidirectional drivers of bias in this study.

Conclusion

We estimated that >3600 cancer cases (3.5% of all incident cancers) in 2015 among Texas adults aged ≥25 were attributable to oncogenic infections. Furthermore, Hispanic (5.6%) and non-Hispanic Black (5.4%) adults had numerically more infection-attributable cancers than non-Hispanic White adults (2.3%). HPV caused the greatest proportion of cancer cases (1.8%), followed by chronic HCV (0.8%), chronic H pylori (0.5%), chronic HBV (0.3%), and HHV-8 (0.1%). Hispanic adults had relatively more cases of HPV- and H pylori–attributable cancers than non-Hispanic White and non-Hispanic Black adults, whereas non-Hispanic Black adults had relatively more cases of cancers caused by HCV, HBV, and HHV-8 than non-Hispanic White and Hispanic adults. These infections are potentially preventable through education, behavioral modification, or vaccination; consequently, their associated cancers may also be prevented. Screening or early diagnosis and treatment may also be helpful in preventing HPV-, HBV-, HCV-, and H pylori–attributable cancers. Thus, cancer prevention programs should continue to integrate infection prevention and control methods.

In Texas, more effort should be made to emphasize the relationship between oncogenic infections and cancer development, especially through public education. Although the Texas Department of State Health Services supports vaccination to prevent HBV and HPV infection and provides online resources that stress the importance of behavioral modification to prevent HBV, HPV, and HCV infection, the link between infection and cancer is not highlighted.²⁵ Moreover, although the Texas Department of State Health Services has published reports on racial/ethnic disparities in cancer health,²⁶ it is unclear whether and how this information has been used in public health practice.

Because racial/ethnic groups may be affected unequally by infection-attributable cancers, prevention programs should incorporate targeted health promotion efforts that account for racial/ethnic differences. For example, because our results suggest that Hispanic adults in Texas may be differentially affected by HPV-associated cancers, striving to improve vaccination compliance and risk perception among largely Hispanic communities in Texas may help reduce cancer disparities.

ORCID iDs

Franciska J. Gudenkauf, MPH https://orcid.org/0000-0002-7463-8934

Aaron P. Thrift, PhD https://orcid.org/0000-0002-0084-5308