Potential role of human papillomavirus in the development of subsequent primary in situ and invasive cancers among cervical cancer survivors

Abstract

BACKGROUND.

The recent licensure of human papillomavirus (HPV) vaccines will likely decrease the development of primary in situ and invasive cervical cancers and possibly other HPV-associated cancers such as vaginal, vulvar, and anal cancers. Because the HPV vaccine has the ability to impact the development of >1 HPV-associated cancer in the same individual, the risk of developing subsequent primary cancers among cervical cancer survivors was examined.

METHODS.

Using the 1992 through 2004 data from the Surveillance, Epidemiology, and End Results (SEER) program, 23,509 cervical cancer survivors were followed (mean of 4.8 person-years) for the development of subsequent primary cancers. The observed number (O) of subsequent cancers of all sites were compared with those expected (E) based on age-/race-/year-/site-specific rates in the SEER population. Standardized incidence ratios (SIRs = O/E) were considered statistically significant if they differed from 1, with an α level of 0.05.

RESULTS.

Among cervical cancer index cases, there was a significant elevated risk for subsequent in situ cancers of the vagina and vulva (SIRs of 53.8 and 6.6, respectively); and invasive vaginal, vulvar, and rectal cancers (SIRs of 29.9, 5.7, and 2.2, respectively). Significantly elevated risks were observed across race and ethnic populations for subsequent vaginal in situ (SIR for whites of 49.4; blacks, 52.8; Asian/Pacific Islander [API], 91.4; and Hispanics, 55.7) and invasive cancers (SIR for whites of 25.7; blacks, 34.5; API, 48.5; and Hispanics, 25.2).

CONCLUSIONS.

The results of the current study demonstrate a substantially increased risk of the development of subsequent primary in situ and invasive cancers among cervical cancer survivors and have implications for the development of prevention and early detection strategies as the role of HPV infection becomes evident. Cancer 2008;113(10 suppl):2919–25. Published 2008 by the American Cancer Society.

Based on recent estimates, there are approximately 250,000 cervical cancer survivors in the US.1 The role of human papillomavirus (HPV) in the development of cervical cancer has been well established.2 A comprehensive study on the global burden of HPV-associated cancers attributes HPV infection to 100% of all cervical cancers.3 Other risk factors such as tobacco use, diet, reproductive factors, and hormones have also been implicated in the development of cervical cancer in various studies.4 Prior studies have shown that women with cervical cancer have increased risk of developing subsequent primary cancers of the vagina, vulva, rectum, anus, ovary, kidney, bladder, breast, and cancers of the aerodigestive tract (lung, bronchus, buccal cavity, pharynx, esophagus, and stomach).5‒10 Some of these primary cancers may have coexisted before the diagnosis of cervical cancer and/or have been diagnosed subsequently or appeared after treatment of cervical cancer with radiotherapy. However, examination of the potential role of HPV in characterizing the burden and risk of developing a subsequent primary cancer among women with cervical cancer has been limited, particularly the risk of subsequent in situ cancers among minorities, especially Hispanics and Asian/Pacific Islanders (API).

The recent introduction of the quadrivalent HPV vaccine, which is effective against HPV types 16–associated and 18-associated cervical, vaginal, and vulvar precancers,11, 12 will likely decrease the development of primary in situ and invasive cervical cancers and possibly other HPV-associated cancers. Currently, the vaccine is approved for girls and women ages 9 through 26 years, not cancer survivors. However, both the quadrivalent and bivalent HPV vaccine manufacturers are seeking approval for the vaccine in older women (ages 25-55 years). By preventing infection with the vaccine HPV types, the vaccine is potentially preventing additional HPV-associated cancers in a woman who would have developed cervical cancer. In the current study, we attempted to quantify the burden of additional HPV-associated cancers among cervical cancer survivors. We examined the risk of developing subsequent primary cancers (in situ or invasive) among cervical cancer survivors using the population-based cancer registry data from the Surveillance, Epidemiology, and End Results (SEER) program. We also examined the risk of subsequent primary cancers among different racial and ethnic groups because Hispanic and African American women have the highest burden of cervical cancer. Detailed examination of their risks could be useful in prioritizing future efforts to increase access to HPV vaccine despite limited financial resources or limited access to healthcare in rural areas.13

MATERIALS AND METHODS

Data

We analyzed cancer incidence data from 1992 through 2004 from 13 population-based cancer registries participating in the National Cancer Institute's SEER program, covering 14% of the US population. These registries are Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco-Oakland, Los Angeles, San Jose-Monterey, Seattle-Puget Sound, Utah, Alaska Native Tumor Registry, and 10 counties in rural Georgia. SEER data regarding expanded race and ethnicity were available beginning in 1992. In addition to patients' demographic characteristics, the data include tumor characteristics and initial therapy. The primary data sources include hospital medical records, pathology and radiotherapy reports, outpatient surgical center records, and death certificates.

Index Cases

Index cases of primary invasive cervical cancer were defined as those diagnosed during the period 1992 through 2004 with no prior reported cancers of any site. Potential index cases in which cervical cancer was not the first diagnosed cancer were excluded, to exclude cancers in which treatment for cancers other than cervical cancer may have contributed to the development of subsequent cancers. This gave us a cohort of index cases in which cervical cancer was the first cancer. We also excluded index cases identified by death certificate or autopsy only because these cases, by definition, do not have follow-up information. Microscopic confirmation rates were >97% for the eligible index cases of cervical cancer.

Subsequent Primary Cancers

We defined subsequent primary cancers (in situ and invasive) as those reported ≥2 months after the index diagnosis date. To reduce the possibility of artifactual association because of intensive screening of cancer patients during the initial medical workup, we excluded cancers diagnosed during the first 2 months of follow-up.14 The subsequent primary cancers included those reported to the SEER incidence registry as well as death certificate–only or autopsy-only cases. Detailed guidelines regarding identifying subsequent primary cancers are provided in the SEER program code manual.14 SEER summary stage was used for classifying stage at diagnosis. The methods overview article by Watson et al15 reported that the SEER summary stage is compatible across the years the data were examined. In an effort to assess the potential effects of treatment (radiotherapy) on the subsequent development of primary cancers, index cases were stratified by mode of treatment and assessed for the risk of developing subsequent primary cancers. The risk of the subsequent development of potentially HPV-associated (vagina, vulva, cervix, rectum, anus, and oropharynx) and non-HPV-associated cancers (esophagus, stomach, colon, corpus uteri, ovary, kidney, bladder, breast, melanoma of skin, and lung and bronchus) were assessed. Because of the relatively small number of subsequent primary cancers (in situ or invasive), we used the SEER site recode International Classification of Diseases for Oncology (ICD-O-3) definition for cancers of the oropharynx.15 In addition, we did not restrict analysis to specific histology subtypes most likely associated with HPV (eg, squamous cell carcinomas of vagina, vulva, rectum, anus, and oral cavity and pharynx) but this was done in most of the analyses reported in this supplement when we assessed for subsequent primary cancers to manage the problem of small numbers. In addition, the risk of subsequent development of HPV-associated cancers was assessed by race and ethnicity.

Statistical Analysis

The analysis was conducted using the person-years and indirect standardization methods. Person-years (PY) at risk for the development of subsequent cancers for each woman began at 2 months after the date of diagnosis of the index cancer and ended at the date of last contact with the patient, death, or end of the study period (December 31, 2004), whichever was earlier. All subsequent cancers (eg, second, third, fourth) for a given case were included. The PY and observed cases of subsequent cancers were stratified according to patient age at initial diagnosis (5-year groups), race (or ethnicity), and 5-year calendar year intervals. Cancer incidence rates among the female population were calculated for the 13 SEER areas combined by age, race, and calendar-year group, and were multiplied by the PY accrued by the index cases to estimate the expected numbers of subsequent cancers for each stratum. The observed and expected numbers of subsequent cancers for each stratum were then summed. The standardized incidence ratio (SIR) represents the ratio of the observed number divided by the expected number of subsequent cancers (O/E). The SIRs describe the risk of development of subsequent primary cancers. To assess the actual burden of subsequent cancers, the absolute excess risk per 10,000 PY (AER) was computed using the formula ((O – E) ÷ PY) × 10,000. The AERs describe the actual burden of subsequent primary cancer in the population.17 Tests on the statistical significance of the SIRs were performed on the basis of the assumptions that O followed a Poisson distribution, and that no variation was associated with E. SIRs and AERs were calculated for both in situ and invasive cancers, respectively. SEER*STAT software was used for data analysis.18

Bidirectional Association

A unidirectional association between cancer of the cervix and a subsequent cancer at a given site may suggest an artifactual association or a treatment effect. A bidirectional association between 2 cancers suggests shared risk factors (genetic or environmental) between the 2 cancers.17 To assess bidirectional relationships, we determined associations between first primary invasive cancers at noncervical sites and subsequent primary invasive cervical cancer (reported ≥2 months after primary cancers at the noncervical sites).

RESULTS

Characteristics of Index Cases

We followed 23,509 index cases of invasive cervical cancer identified in the SEER registries for a mean of 4.8 PY. Among the index cases, 74% were white, 13% were black, and 11% were API (Table 1). Approximately 23% of the index cases were Hispanic. Approximately 48% of patients were ages 30 to 49 years at the time of diagnosis. The mean age at diagnosis was 47 years. Squamous cell (69%) and adenocarcinoma (18%) constituted the major histologic types. The majority of the index cases were either localized (52%) or regional stage (33%) at the time of diagnosis. Approximately 47% of the index cases were reported to have received radiation as a mode of treatment.

Table 1. Characteristics of Index Cases of Invasive Cervical Cancer, SEER Program 1992-2004

Characteristic	Total N=23,509 % (No)
SEER indicates Surveillance, Epidemiology, and End Results program; API, Asian Pacific Islander. * Microscopically confirmed. Data were obtained from 13 population-based cancer registries that participate in the SEER program and meet high-quality data criteria. These registries cover approximately 14% of the US population.
Race
White	74.1 (17,423)
Black	12.9 (3039)
API	11.0 (2576)
Other	2.0 (471)
Ethnicity
Non-Hispanic	76.9 (18,072)
Hispanic	23.1 (5437)
Age at diagnosis, y
Mean age at diagnosis (age range)	47 (6-104)
<30	6.8 (1593)
30-49	48.5 (11,410)
50-69	29.9 (7034)
≥70	14.8 (3472)
Histology*
Squamous cell and transitional carcinoma	69.4 (16,320)
Adenocarcinomas	18.4 (4330)
Adenosquamous and glassy cell	4.8 (1122)
Small cell/neuroendocrine	1.0 (257)
Other carcinomas (specified and unspecified)	3.6 (846)
Noncarcinomas	1.4 (331)
Stage of diagnosis
Localized	51.6 (12,142)
Regional	33.0 (7748)
Distant	8.8 (2071)
Unstaged	6.6 (1548)
Type of treatment
Radiation	47.0 (11,043)
No radiation	51.1 (12,023)
Radiation status unknown	1.9 (443)
Mean person-years at risk	4.8

Risk and Burden of Subsequent Invasive Cancers

During the follow-up period, 902 subsequent invasive cancers were identified (Table 2). The HPV-associated subsequent invasive cancers accounted for 13% (N = 115) of all subsequent invasive cancers. The SIRs over the total follow-up period were significantly elevated for HPV-associated invasive cancers of the vagina, vulva, and rectum (SIRs of 29.87, 5.74, and 2.16, respectively) (Table 2). The AER of subsequent invasive cancers of the vagina, vulva, and rectum were 3.27, 1.52, and 1.38 per 10,000 PY, respectively. Elevated risk was observed for invasive cancer of the oropharynx (SIR of 1.68); however, it was not statistically significant.

Table 2. Subsequent Invasive Cancers Among Women With Invasive Cervical Cancer, All Races, SEER Program 1992-2004

Subsequent Primary Invasive Cancer	Follow-up Time Since Diagnosis
	2-59 months		≥60-156 months		Total(≥2 months)
	O	SIR	O	SIR	O	SIR	AER
SEER indicates Surveillance, Epidemiology, and End Results program; O, observed cancers; SIR, standardized incidence ratio; AER, Absolute Excess Risk; HPV, human papillomavirus. * SIR differs significantly from 1 (P < .05). Data were obtained from 13 population-based cancer registries that participate in the SEER program and meet high-quality data criteria. These registries cover approximately 14% of the US population. Dashes indicate that data were suppressed if there were <6 observed cancers.
All sites	626	1.60*	276	1.24*	902	1.47*	29.43
HPV-associated cancers
Oral cavity and pharynx	12	1.96*	–	–	16	1.68	0.66
Rectum	17	2.33*	8	1.87	25	2.16*	1.38
Anus	–	–	–	–	–	–	–
Cervix	16	1.63	–	–	20	1.38	0.56
Vagina	29	41.29*	–	–	33	29.87*	3.27
Vulva	12	6.02*	6	5.25*	18	5.74*	1.52
Non HPV-associated cancers
Esophagus	7	3.64*	–	–	8	2.66*	0.51
Stomach	11	1.79	–	–	15	1.56	0.55
Colon	40	1.29	23	1.31	63	1.3	1.49
Corpus uteri	18	0.77	13	0.96	31	0.84	−0.62
Ovary	25	1.88*	7	0.93	32	1.54*	1.15
Kidney	13	1.89*	–	–	17	1.53	0.6
Bladder	35	4.37*	–	–	40	3.16*	2.81
Breast	115	0.87	56	0.74*	171	0.82*	−3.73
Melanoma of skin	11	0.85	–	–	15	0.73	−0.56
Lung and bronchus	150	3.27*	70	2.66*	220	3.04*	15.16

For non-HPV-associated cancers, significantly elevated SIRs were observed for subsequent invasive cancers of esophagus, ovary, bladder, and lung and bronchus. A significant decreased risk was found for cancer of the breast (SIR of 0.82; P < .05). The AER for subsequent invasive cancer of the lung and bronchus was 15.16 per 10,000 PY, whereas AERs for subsequent cancers of the esophagus, ovary, and bladder were 0.51, 1.15, and 2.81 per 10,000 PY, respectively.

For HPV-associated cancers that demonstrated significantly elevated risks, we examined the risk by race and ethnicity (Fig. 1). The SIR of subsequent invasive cancers of the vagina was found to be significantly elevated among all racial and ethnic groups (SIRs for whites of 25.66; blacks, 34.54; API, 48.48; and Hispanics, 25.15) (Fig. 1). The SIR of developing subsequent vulvar cancers (SIRs for whites of 6.49, and 12.78 for Hispanics) and rectal cancers (SIRs for whites of 2.16, and 3.33 for Hispanics) were found to be significantly elevated among whites and Hispanics; the associations could not be assessed for blacks and API because of small numbers. The AER was found to be substantially elevated for subsequent invasive cancers of the vagina among all racial and ethnic groups.

Figure 1. Subsequent invasive cancers of selected sites among women with invasive cervical cancer by race and ethnicity according to the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program, 1992-2004. *P < .05; †data were suppressed if there were <6 observed cancers; ‡Hispanic origin is not mutually exclusive from race categories (White, Black, Asian/Pacific Islander [API]).

We also examined HPV-associated cancers that demonstrated significantly elevated risks by age at diagnosis and by mode of treatment (Table 3). The SIR was higher for subsequent invasive cancers of the vagina among women ages 30 to 49 years compared with women ages 50 to 69 years (Table 3). However, the burden of subsequent vaginal cancer, as assessed by AER, was highest among the oldest women. A similar pattern of higher burden with increasing age at diagnosis of the index cancer was observed for subsequent cancers of the vulva and rectum. Significantly elevated SIRs were found for subsequent vaginal cancers among women irrespective of whether they received radiotherapy for treatment of the index cancer. In addition, significantly elevated risks were found for subsequent vulvar and rectal cancers among women who received radiotherapy (SIRs of 10.49 and 2.55, respectively).

Table 3. Subsequent Invasive Cancers Among Women With Invasive Cervical Cancer by Age at Diagnosis and Mode of Treatment, SEER Program 1992-2004

Characteristics of Index Case	Subsequent Primary Cancer
	Vaginal			Vulval			Rectum
	O	SIR	AER	O	SIR	AER	O	SIR	AER
SEER indicates Surveillance, Epidemiology, and End Results program; O, observed cancers; SIR, standardized incidence ratio; AER, Absolute Excess Risk; HPV, human papillomavirus. * SIR differs significantly from 1 (P < .05). Data were obtained from 13 population-based cancer registries that participate in the SEER program and meet high-quality data criteria. These registries cover approximately 14% of the US population. Dashes indicate that data were suppressed if there were <6 observed cancers.
Age at diagnosis
<30	0	–	–	0	–	–	0	–	–
30-49	15	51.92*	2.70	–	–	–	7	2.57*	0.79
50-69	15	28.11*	5.54	7	5.81*	2.22	12	2.31*	2.60
≥70	–	–	–	6	5.90*	6.11	6	1.77	3.20
Mode of treatment
No radiation	16	29.66*	2.6	–	–	–	10	1.78	0.74
Radiation	17	30.38*	4.39	16	10.49*	3.86	15	2.55*	2.43

Risk and Burden of Subsequent In Situ Cancers

We identified 186 subsequent in situ cancers during the follow-up period. The SIRs were found to be significantly elevated for subsequent in situ vaginal and vulvar cancers (SIRs for vaginal cancer of 53.8 and 6.6 for vulvar cancer, respectively) (Fig. 2). The AERs for subsequent in situ vaginal and vulvar cancers were 6.7 and 3.3, per 10,000 PY, respectively. The significant elevated risk for in situ vaginal cancer alone was observed by race and ethnicity (SIRs for whites of 49.4; blacks, 52.8; API, 91.4; and Hispanics, 55.7, respectively). Due to the small number of subsequent in situ cancers, we were unable to assess the risk for other HPV-associated cancers.

Figure 2. Subsequent in situ cancers of selected sites among women with invasive cervical cancer according to the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program, 1992-2004. *P < .05; †data were suppressed if there were <6 observed cancers.

Bidirectional Association

Bidirectional association was assessed between cancers of the cervix and the other cancer sites (Table 4). Although elevated SIRs (SIRs >1) for subsequent invasive cervical cancer were found among women with cancers of oropharynx and rectum, the association was not statistically significant. Bidirectional associations could not be assessed for cancers of the vagina and vulva because of the small numbers. Significant decreased risks (SIRs <1) were found between first primary cancers of the corpus uteri, breast, melanoma of skin, and lung and bronchus and subsequent cervical cancer.

Table 4. Results From Analysis Assessing the Bidirectional Association (Subsequent Invasive Cervical Cancer by Site of First Cancer), SEER Program, 1992-2004

First Cancer Site	O	SIR	AER
SEER indicates SEER indicates Surveillance, Epidemiology, and End Results program; O, observed cancers, SIR, Standardized Incidence Ratio, AER, Absolute Excess Risk; HPV, human papillomavirus. * SIR differs significantly from 1 (P < .05). Dashes indicate that data were suppressed if there were <6 observed cancers. Data were obtained from 13 population-based cancer registries that participate in the SEER program and meet high-quality data criteria. These registries cover approximately 14% of the US population.
HPV-associated cancers
Oral cavity and pharynx	8	1.19	0.27
Rectum	12	1.45	0.65
Anus	0	–	–
Vagina	–	–	–
Vulva	–	–	–
Non-HPV-associated cancers
Esophagus	–	–	–
Stomach	6	1.73	1.17
Colon	44	1.30	0.43
Corpus uteri	14	0.41*	−0.84
Ovary	–	–	–
Kidney	6	0.81	−0.27
Bladder	11	1.12	0.17
Breast	114	0.59*	−0.59
Melanoma of skin	6	0.31*	−0.92
Lung and bronchus	10	0.51*	−0.73

Sensitivity Analysis

We conducted sensitivity analyseis to assess the impact of change in the index case definition on the risk and burden of development of subsequent primary cancers. We performed 1 sensitivity analysis by expanding the index case definition to include both in situ and invasive cervical cancers. The analysis yielded similar results with the exception that the risk of subsequent vulvar cancer among blacks was found to be significantly elevated (SIR of 13.0; P < .05) (data not shown). In other sensitivity analyses, we considered subsequent cancers, excluding those diagnosed <12 months after the index case and assessed the risk and burden of those who were followed up for ≥12 months; these results were found to be no different from our reported findings.

DISCUSSION

The findings of the current study are consistent with previous studies that demonstrated an elevated risk for subsequent primary invasive cancers among cervical cancer survivors.5–10 However, to our knowledge, this is the first study to report the risk and burden of subsequent in situ cancers among cervical cancer survivors, and also to report the risk and burden of subsequent primary cancers among Hispanics and API women.

The elevated risks for subsequent invasive cancers of the vagina, vulva, and rectum and in situ cancers of the vagina and vulva support a potential etiologic role of HPV. The elevated risks of subsequent cancers observed in the current study are unlikely to be attributable to radiotherapy because radiotherapy-associated cancers generally appear 10 to 15 years after therapy.19 The mean follow-up period in the current study was 4.8 PY. We observed a substantially elevated risk for subsequent in situ and invasive vaginal cancer among white, black, API, and Hispanic women with cervical cancer. It should be noted that a portion of the elevated risk of invasive female genital cancers may reflect missed opportunities to identify these subsequent cancers at the earlier, in situ stage because of inadequate healthcare access and follow-up.20, 21 Nevertheless, these findings could help formulate preventive strategies such as closer monitoring (post–treatment surveillance), vigilant screening for HPV-associated cancers and other cancers, and health education for risk factor reduction (such as tobacco use) among cervical cancer survivors to improve survivorship.

Routine screening of women for vaginal cancer after a hysterectomy for benign disease is not recommended, but the current study data demonstrate that women with a history of high-grade cervical intraepithelial neoplasia or invasive cervical cancer are at increased risk and may benefit from a regular vaginal cuff Papanicolaou (Pap) test. Vaginal colposcopy is also indicated in the presence of concomitant cervical and vulvar lesions. To our knowledge there is no recommended screening procedure for vulvar cancer. However, for women with a past history of cervical or vaginal intraepithelial neoplasia/cancer, an inspection of the vulva with or without colposcopy could be encouraged as part of their regular follow-up. We also found that women with cervical cancer are at increased risk for developing the several tobacco-related cancers (esophagus, bladder, and lung and bronchus), which is likely because of tobacco use among women with cervical cancer as prior studies have suggested.22, 23 Because smoking is a known risk for cervical cancer, women diagnosed with cervical cancer should be targeted for tobacco cessation interventions. Often a cancer diagnosis presents a ‘teachable moment’ in which tailored advice could result in positive health behaviors.24, 25

There are a few potential limitations with this study. First, the small numbers of subsequent primary cancers limited further subgroups analysis, particularly by histology among HPV-associated and non-HPV-associated cancers. Second, it is possible that some of the cervical cancer patients had moved out of the SEER areas, resulting in underreporting of subsequent cancers. Finally, the role of detection bias in the diagnoses of some of subsequent cancers cannot be completely eliminated.

The study findings shed new light on the potential role of HPV in the development of subsequent primary cancers among cervical cancer survivors, particularly the in situ (vaginal and vulvar) cancers. The findings could help formulate preventive strategies such as post–treatment surveillance, screening, and risk factor reduction education among cervical cancer survivors to improve survivorship. The advent of the HPV vaccine and its efficacy in preventing cervical cancer has given us hope in the fight against cancer. Although the role of currently licensed HPV vaccine is preventive in nature, knowledge of the additional burden that the same individual may face may be an important factor to consider in future research.