Medical Oncology

, 31:943

Risk of colorectal cancer among long-term cervical cancer survivors

Authors

    • Department of Obstetrics and GynecologyThe University of Texas Medical Branch
  • Yong-Fang Kuo
    • Department of Internal Medicine, Sealy Center on AgingThe University of Texas Medical Branch at Galveston
  • James S. Goodwin
    • Department of Internal Medicine, Sealy Center on AgingThe University of Texas Medical Branch at Galveston
Original Paper

DOI: 10.1007/s12032-014-0943-2

Cite this article as:
Rodriguez, A.M., Kuo, Y. & Goodwin, J.S. Med Oncol (2014) 31: 943. doi:10.1007/s12032-014-0943-2

Abstract

Because advances in therapy have increased long-term survival for women with cervical cancer, it is important to study the risk of secondary primary malignancies in high-dose organ areas. From the 1973–2009 National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) program, we studied the risk of developing cancer of the colon and rectum in 64,507 cervical cancer patients over 35 years after initial radiation treatment. We also assessed change in risk over time. Kaplan–Meier estimator for survival curve and Cox proportional hazards models was used. More than half (52.6 %) of the cervical cancer patients received radiation treatment. In the analyses adjusted for race/ethnicity, age, marital status, surgery status, stage and grade, the risk of colon cancer between those both with and without XRT diverged beginning at approximately 8 years. After 8 years, the hazard ratio for developing colon cancer was 2.00 (95 % CI 1.43–2.80) for women with radiation versus those without radiation treatment. The risk of rectal cancer diverged after 15 years of follow-up (HR 4.04, 95 % CI 2.08–7.86). After 35 years of follow-up, the absolute risk of developing colon cancer was 6.5 % for those who received radiation versus 2.5 % for those without, and 3.7 versus 0.8 % for rectum. The risk of colon and rectum cancer over 20 years of follow-up after radiation remained the same across three eras (1973–1980, 1981–1990, and 1991–2000). Radiation-induced second cancers of the colon and rectum may occur 8 years after radiation treatment for cervical cancer.

Keywords

Cervical cancerSEERSecond primary cancerRadiotherapyColon cancerRectal cancer

Introduction

The incidence and mortality for cervical cancer has decreased in the United States [1, 2]. As of 2012, of the 245,022 women surviving cervical cancer, 83 % have already lived 5 years or more after the diagnosis [3, 4]. With the growing number of people surviving cancer in general, it is vital to study the health complications that might develop during the months or years following the completion of the given treatment to treat the primary tumor [5]. Developing a second cancer is one of the most serious complications of cancer treatment [6, 7]. An estimated 18 % of the incident malignancies in the United States are a second (subsequent) cancer [8].

Patients with cervical cancer provide an excellent opportunity to study the lasting effects of radiotherapy [6]. There is a sufficient number of patients available to study, including non-irradiated patients available for comparison. In addition, radiation doses received by other organs while treating cervical cancer can be estimated accurately [6, 9]. These doses range from tens of grays (thousands of rads) for those organs nearest to the cervix to tens of grays (tens of rads) for those farthest away [6]. Organs located in the pelvis including the rectum have been determined to receive very high doses of radiation (>30 Gy) while colon was found to receive varied doses, ranging from 4 Gy for the transverse colon to 24 Gy for the sigmoid colon [6].

The choice of radiotherapy is dependent on the stage of disease [10]. Radiation treatment consists of external beam radiotherapy, brachytherapy using an intracavitary radiation source, such as radium or cesium or a combination of both modalities [10].

Previous studies have established that cervical cancer survivors treated with radiation have an increased risk for the development of second primary malignancies of the colon, rectum/anus, bladder and ovary [1114]. However, no recommendations have been established in terms of which specific screening strategies should be implemented in this group of women or how those findings can be applied to women currently surviving cervical cancer.

The aim of this study was to assess the risk of developing SPMs of colon and rectum over 35 years of follow-up after radiation treatment for cervical cancer and whether this risk has decreased over time as methodologies for radiation therapy have evolved.

Materials and methods

Data source

Data for this study came from the National Cancer Institute’s Surveillance, Epidemiology and End Results (SEER) program. The SEER program has been collecting data on cancer incidence and survival since 1973 and publishes cancer statistics for approximately 28 % of the United States population [15]. This program currently has 18 population-based cancer registries across the United States and covers 25 % of non-Hispanic Whites, 26 % of African Americans, 38 % of Hispanics, 110.3 % American Indian/Alaskan Native/Pacific Islander and 50.4 % other groups [15].

The SEER Program is an excellent resource to study cancer incidence and second malignancies in the United States with case ascertainment rate at 97.5 % [16]. All cancer sites are categorized using the International Classification of Diseases for Oncology, 3rd Edition (ICD-0-3). Institutional review board approval for analyses was deemed not necessary because all identifying information had been removed.

Study cohort

Initially, a total of 68,563 women with stage I–IV cervical cancer (C530–C539) were identified from 1973 to 2009 in the SEER database. Patients were excluded if they had been diagnosed by autopsy or death certificate only (n = 28), were younger than 20 years of age or older than 85 years of age (n = 1,430) and had unknown radiation (n = 886) or unknown surgical status (n = 1,712). After applying all the exclusion criteria, the final sample included 64,507 females.

Study variables

The study variables included race/ethnicity, age at diagnosis, marital status, radiation status (yes/no), surgery status (yes/no), tumor stage (SEER historic stage A), tumor grade, SPMs in colon and rectum, and time to develop SPMs on colon and rectum. Racial/ethnic groups were defined as non-Hispanic (NH) White, NH Black, Hispanic, American Indian/Alaskan Native/Pacific Islander and other. Age at diagnosis is the age (in years) that the patient was diagnosed with cervical cancer and was categorized into six categories (20–30, 31–40, 41–50, 51–60, 61–70, >71). Marital status was dichotomized into married and not married (single (never married), separated, divorced or widowed). Tumor stage was categorized as localized, regional or distant. Tumor grade is based on the International Classification of Diseases for Oncology, 2nd Edition (ICD-0-2) and categorized as well differentiated, moderately differentiated, poorly differentiated, undifferentiated or cell type not determined, not stated or not applicable. Cancers of the colon (C180–C189, C260) and the rectum (C199, C209) were identified by linking registry records based on subject numbers. The study outcome, time to develop cancers of the colon and/or rectum, was calculated from the time of initial diagnosis of cervical cancer to when the malignancy developed on the colon and rectum. Patients were censored at death or at the end of the follow-up period (December 31, 2009).

Statistical analysis

Chi-square tests were used to compare patient and tumor characteristics between patients who did or did not receive radiation treatment for cervical cancer. Risk of subsequent colon and rectal cancer was estimated by the Kaplan–Meier method and compared by log-rank test for patients with and without radiation treatment. Cox proportional hazards models were developed to estimate the risk, and proportional hazards assumption was examined by hazard plot and Sheffield residuals [17]. When the proportional hazards assumption was violated—that is, the hazards were not proportional over the 35 years of follow-up—the time-dependent effect of radiation up to a specific year and after that year was included in the model. The location of the knot point at the specific year of follow-up was estimated by nonlinear least squares regression on the differences of hazard between two groups over time. To assess whether the risk associated with radiation changed over the past three decades, we limited the follow-up time to 20 years and grouped patients diagnosed with colorectal cancer into four groups (1973–1980, 1981–1990, and 1991–2000). We tested the interaction between diagnosed year group and radiation in the Cox proportional hazards model. We also explored whether the risk of colorectal cancer associated with radiation varied by patient and cancer characteristics by testing their interactions in the Cox proportional hazards models. All statistical analyses were performed using SAS software (Version 9.3, SAS Institute, and Cary, NC, USA). All p values are two-tailed with pvalue ≤ 0.05 considered statistically significant.

Results

Demographics, clinical and tumor characteristics

From 1973 to 2009, there were 64,507 incident cases of cervical cancer. A summary of demographics, tumor characteristics and treatment is shown in Table 1. More than half (52.6 %) of the cohort received radiation therapy as treatment. There were significant differences in race/ethnicity, age, marital status, surgery status, stage and grade between the radiation and no radiation groups. Compared to the radiation group, the no radiation group had more localized stage disease, was younger and had a higher percentage of surgeries. The no radiation group also had a higher percentage of not being married (51.5 vs 44.3 %) and unknown tumor grade (49.2 vs 29.4 %). Figure 1 shows the proportion of women who received radiation treatment by year. The percent receiving radiation therapy initially decreased, from 1973 to 1999. It increased from 2000 to 2005. Since 2006, the proportion of women receiving radiation therapy has slightly decreased.
Table 1

Demographics, pathologic and surgical treatment characteristics between patients with cervical cancer who received radiation and patients who did not receive radiation, SEER 1973–2009 (n = 64,507)

Variables

Radiation (n = 33,961)

No radiation (n = 30,546)

p value

n

%

n

%

Race/ethnicity

<0.0001

 Non-Hispanic white

19,106

56.3

18,668

61.1

 

 Non-Hispanic black

5,459

16.1

3,445

11.3

 Hispanic

6,099

18.0

5,379

17.6

 AAPI

2,836

8.4

2,425

7.9

 Other

461

1.4

629

2.1

Age at diagnosis

<0.0001

 20–30

1,443

4.2

4,215

13.8

 

 31–40

5,782

17.0

9,392

30.7

 41–50

8,418

24.8

7,573

24.8

 51–60

7,517

22.1

4,217

13.8

 61–70

5,860

17.3

2,924

9.6

 ≥71

4,941

14.5

2,225

7.3

Mean age at cervical cancer diagnosis (years ± SD)

53.3 ± 14.5

45.3 ± 14.4

 

Marital status

<0.0001

 Married

17,831

52.5

13,131

43.0

 

 Not married

15,050

44.3

15,740

51.5

 Unknown

1,080

3.2

1,675

5.5

Surgery

<0.0001

 Yes

13,545

39.9

26,809

87.8

 

 No

20,416

60.1

3,737

12.2

Stage

<0.0001

 Localized

9,729

28.6

25,611

83.8

 

 Regional

19,889

58.6

3,035

9.9

 Distant

4,343

12.8

1,900

6.2

Grade

<0.0001

 Well differentiated

1,963

5.8

3,220

10.5

 

 Moderately differentiated

9,445

27.8

6,392

20.9

 Poorly differentiated

11,489

33.8

5,331

17.5

 Undifferentiated

1,090

3.2

587

1.9

 Not determined

9,974

29.4

15,016

49.2

https://static-content.springer.com/image/art%3A10.1007%2Fs12032-014-0943-2/MediaObjects/12032_2014_943_Fig1_HTML.gif
Fig. 1

Proportion of cervical cancer patients receiving radiation after diagnosis by study year, SEER 1973–2009 (n = 64,507)

Second primary malignancies of colon and rectum after radiation treatment for cervical cancer

Figures 2 and 3 demonstrated the risk of a second cancer at the colon or rectum over time estimated by the Kaplan–Meier method. The overall risk at 35 years of follow-up for cervical cancer patients receiving radiation therapy was 6.5 % for colon cancer and 3.7 % for rectal cancer. In comparison, cervical cancer patients without radiation therapy had a 2.5 % risk for colon and 0.8 % risk for rectal cancer.
https://static-content.springer.com/image/art%3A10.1007%2Fs12032-014-0943-2/MediaObjects/12032_2014_943_Fig2_HTML.gif
Fig. 2

Kaplan–Meier estimates for colon cancer for patients with cervical cancer who did or did not receive radiation treatment

https://static-content.springer.com/image/art%3A10.1007%2Fs12032-014-0943-2/MediaObjects/12032_2014_943_Fig3_HTML.gif
Fig. 3

Kaplan–Meier estimates for rectal and anal cancer for patients with cervical cancer who did or did not receive radiation treatment

Table 2 shows results from the Cox proportional hazards models to further assess the effect of radiation adjusted for age, race/ethnicity, marital status, year of diagnosis, surgical treatment and tumor characteristics. We found that the hazards of colon and rectal cancer associated with radiation were not proportional over time (p < 0.001). Therefore, we performed nonlinear least square regression to find the optimal cutoff on the follow-up time associated with change of hazard. For colon cancer, there were no statistically significant differences seen between women with and without radiation within 8 years of follow-up. However, at more than 8 years of follow-up, the hazard ratio for colon cancer was 2.00 (95 % CI 1.43–2.80) for women with radiation compared to those who did not receive radiation treatment. The risk of rectal cancer became significant at more than 15 years of follow-up (HR 4.04, 95 % CI 2.08–7.86).
Table 2

Proportional hazards to colon, rectal and anal cancer for cervical cancer patients (n = 64,507)

Characteristic

Category

Colon cancer*

Rectum**

HR

95 %

CI

HR

95 %

CI

Radiation

 ≤8 years

No

1.00

     

Yes

1.13

0.81

1.56

   

 >8 years

No

1.00

     

Yes

2.00

1.43

2.80

   

 ≤15 years

No

   

1.00

  

Yes

   

1.37

0.90

2.10

 >15 years

No

   

1.00

  

Yes

   

4.04

2.08

7.86

Age (every 5 years)

 

1.38

1.33

1.43

1.34

1.25

1.43

Race/ethnicity

White

1.00

  

1.00

  

Black

1.23

0.94

1.62

1.31

0.86

1.99

Hispanic

0.60

0.41

0.89

0.81

0.48

1.36

Others

0.68

0.47

0.98

1.14

0.72

1.81

Marital statusa

No

1.00

  

1.00

  

Yes

1.22

0.99

1.51

1.08

0.79

1.48

Stage

Local

1.00

  

1.00

  

Regional

0.96

0.76

1.22

1.08

0.75

1.55

Distant

1.11

0.65

1.92

0.54

0.17

1.75

Grade

Well differentiated

1.00

  

1.00

  

Moderately differentiated

1.29

0.85

1.95

0.98

0.57

1.70

Poorly differentiated

1.17

0.77

1.78

0.62

0.34

1.13

Undifferentiated

1.43

0.73

2.8

0.43

0.10

1.85

Not determined

0.98

0.66

1.46

0.80

0.48

1.35

Year of diagnosis

 

0.99

0.98

1.01

1.02

0.99

1.04

Surgery

No

1.00

  

1.00

  

Yes

0.94

0.74

1.20

1.35

0.92

1.96

a Unknown was not reported

* None of interactions between radiation and covariates were significant (all p value >0.19)

** None of interactions between radiation and covariates were significant (all p value >0.23)

There was no significant interaction between year of diagnosis and radiation treatment on the risk of malignancies at colon and rectum (p = 0.5123, 0.3769, respectively). We also did not find any interactions between age, ethnicity or tumor characteristics, with an increased risk of colorectal cancer associated with radiation (all p values >0.19 for interactions of colon cancer and >0.23 for interactions of rectal cancer).

Discussion

The primary purpose for conducting this study was to evaluate the risk of second cancer in the colon and rectum among long-term survivors of cervical cancer who received radiation therapy. Our study results show that tumors of the colon, rectum and anus are 2–4 times more frequent in the radiation group than in the no radiation group among cervical cancer survivors. The relationship between therapeutic radiation and the risk of SPMs has long been recognized [13].

Previous research has shown that the risk of developing cancer in organs close to the cervix (i.e., bladder, rectum, uterine corpus, ovary, small intestine, bone and connective tissue) that had high radiation exposures since treatment increased with time [18]. The results of this study are consistent with past studies which have shown a positive dose–response relationship for second cancers of the bladder, rectum, bone and stomach following radiation treatment for cervical cancer [13, 19].

Based on our findings, we believe that a woman who developed cervical cancer and received radiation at a young age should not wait to start screening for colorectal cancer at the age currently recommended for low-risk women. Such women should start screening at approximately 8 years after treatment if cervical cancer occurs before the woman reaches the age of 50, the usual age to initiate screening in average-risk individuals. Also, the higher rates of colorectal cancer among these women indicate that more frequent screening for colorectal cancer should be considered.

The results should be interpreted with the recognition of the strengths and weaknesses of cancer registry-based data. One is the lack of information regarding lifestyle choices, such as smoking. Although radiotherapy is commonly noted as a major causative factor in the development of SPMs, <10 % of SPMs are due to radiotherapy, with most attributable to lifestyle and genetics [13]. Smoking can exert variable effects on the risk of gynecologic cancer [2023]. Specifically, cigarette smoking can increase the risk of SPMs in cervical cancer survivors [6, 18]. Cigarette smoking is also a risk factor for colorectal cancer [24]. Since information on lifetime active smoking is not available in this dataset, we are unable to determine the effect modification between colorectal cancer risk and smoking in patients with cervical cancer. A recent study by Hoffmeister et al. [24] found that the efficacy of colorectal screening by endoscopy changes when smoking is discontinued. This suggests that smoking status should be taken into consideration when looking at factors affecting screening and surveillance for colorectal cancer [24, 25]. Another limitation of this study is that second cancers diagnosed among patients who migrate out of their SEER registry area are not reportable, and migration status is not recorded in SEER [15]. This would affect the estimation of absolute risk, but probably not relative risk. Treatment information recorded in cancer registries is not always accurate, and radiation status may have been misclassified, though this is thought to be small [18]. Also, registries focus on documenting treatment initiated within 4–6 months of diagnosis. Subsequent radiation treatment would not be captured [18]. We also could not take into account differences in radiation doses across treatment modalities. An unavoidable limitation of studying the late effects of radiotherapy is that the most recent changes in practice cannot be studied. And lastly, these tumors might not represent second primary malignancies but recurrence or metastasis of the primary cervical tumor.

Despite these limitations, there were strengths to this study. This study included a large cohort size that included all women with cervical cancer since SEER started collecting data in 1973. There was a long follow-up interval, up to 35 years. Patients with cervical cancer provide an excellent opportunity to study the late effects of radiotherapy because of the sufficiently large numbers of patients available for study, patients survive for long periods of time, non-irradiated patients are available for comparison and radiation doses to organs other than the cervix can be estimated accurately [6, 10].

Future endeavors should focus on the creation and implementation of surveillance strategies aimed to the early detection of colorectal tumors among cervical cancer survivors.

Acknowledgments

Sarah Toombs Smith, PhD, Science Editor and Assistant Professor in the Sealy Center on Aging, University of Texas Medical Branch at Galveston, provided editorial assistance in manuscript preparation. Dr. Toombs Smith received no compensation for her assistance apart from her employment at the institution where the study was conducted. Comparative Effectiveness Research on Cancer in Texas (CERCIT) is a statewide resource for outcomes and comparative effectiveness research funded by The Cancer Prevention Research Institute of Texas (CPRIT), RP101207.

Conflict of interest

None.

Copyright information

© Springer Science+Business Media New York 2014