Association of Colorectal Cancer and Prostate Cancer and Impact of Radiation Therapy

To quantify the risk of prostate cancer after colorectal cancer and the risk of colorectal cancer after prostate cancer and to examine the impact of radiation therapy on subsequent cancer risk, we conducted retrospective cohort studies using data from the Surveillance, Epidemiology and End Results program from 1973 to 2005. Standardized incidence ratios (SIR) and 95% confidence intervals (95% CI) were calculated, adjusting for age, ethnicity, and calendar year. The subsequent risk of developing a prostate cancer was significantly elevated in patients diagnosed with colon cancer before age 50 years (SIR, 1.38; 95% CI, 1.18-1.60). The risk of subsequent prostate cancer was decreased for men with rectal cancer who received radiation therapy (SIR, 0.57; 95% CI, 0.52-0.63). Interestingly, this beneficial effect of radiation therapy was only observed in the prostate-specific antigen (PSA) era (1988+). In addition, the prostate cancer cases developed in the radiation therapy group tended to have higher-grade, later-stage tumors, higher PSA levels, and worse survival than those developed in the nonradiation therapy group. In the cohort of prostate cancer patients, the risk of colon cancer was elevated in patients diagnosed with prostate cancer before age 50 years (SIR, 1.51; 95% CI, 1.03-2.20). In conclusion, a diagnosis of colon or prostate cancer in men of younger ages may be an indication for screening of prostate or colon cancer, respectively. The decreased prostate cancer risk in men who received radiation therapy for rectal cancer may be related to the use of PSA for prostate cancer screening or the cure of occult prostate cancer. (Cancer Epidemiol Biomarkers Prev 2009;18(7):1979–85)

Because guidelines for cancer screening are different for increased-risk and average-risk adults, it is important to identify individuals at increased risk compared with the general population (1-3). Identifying extracolonic neoplasms that are associated with the development of colorectal cancer provides important markers for high-risk patients. Several studies have examined the risk of colorectal cancer after the diagnosis of prostate cancer. Most of them showed that there was no overall increased risk but there is a debate on whether the risk of rectal cancer is elevated in the subgroup of patients who received radiation therapy for prostate cancer, even among researchers analyzing the same database (4-14). Limited sample size in some studies and different choices of the comparison groups and analysis methods are possible reasons for this discrepancy.

Studies of the other direction of the association, the risk of subsequent prostate cancer after the diagnosis of colorectal cancer, have also yielded discordant results. Some studies showed an increased risk of prostate cancer after colorectal cancer (15-18), whereas others found no association (19-21). We found that men with rectal cancer who received radiation therapy had a decreased incidence rate of subsequent prostate cancer compared with the general population (22). This finding was further confirmed by two studies that examined the same databases although with different analysis strategies (23, 24). However, the reason for this apparent beneficial effect is unclear and warrants further studies.

In this study, we analyzed the Surveillance, Epidemiology and End Results (SEER) database from 1973 to 2005 to further quantify the risk of subsequent prostate cancer after colorectal diagnosis and the risk of subsequent colorectal cancer after prostate cancer. This study also assessed the impact of radiation therapy for the first primary cancer on the second primary cancer and explored possible reasons for the association.

To have precise estimates of relative risks and sufficient power for subgroup analysis, we used data from all 17 SEER cancer registries, which cover 26% of the U.S. population for the recent years.⁴

Two retrospective cohorts were formed based on the SEER public database. The first cohort consists of men diagnosed with colorectal cancer between 1973 and 2005. We identified 291,823 men diagnosed with colorectal cancer, of which we excluded those in whom a colorectal cancer diagnosis was made at autopsy or by death certificate (n = 2,798); those with no follow-up time (n = 9,462); those whose age, race, or radiation therapy status were unknown (n = 4,746); and those with prostate cancer previously (n = 11,935). After these exclusions, 262,882 patients remained: 177,834 patients with colon cancer and 85,048 patients with rectal cancer. The colon cancer cohort and the rectal cancer cohort were analyzed separately because of etiologic heterogeneity. Using the unique identifier assigned to each patient, the SEER database was searched for subsequent malignant prostate cancer. Follow-up time was defined as time from date of colorectal cancer diagnosis to either the date of diagnosis of the prostate cancer, the date of last follow-up, or the date of death, whichever occurred first.

The second cohort consists of patients with prostate cancer. Of the 677,816 men diagnosed with prostate cancer, we excluded those in whom prostate cancer diagnosis was made at autopsy or death certificate (n = 8,719); those with no follow-up time (n = 9,785); and those whose age, race, or radiation therapy status were unknown (n = 23,402). After these exclusions, 635,910 patients remained in the prostate cancer cohort. Then, we searched for subsequent invasive colon and rectal cancer using the aforementioned method.

The incidence rates of subsequent prostate cancers in the colorectal cancer cohort and the rates of subsequent colorectal cancers in the prostate cancer cohort were then compared with the standard rates in the general population that were also derived from the SEER program from 1973 to 2005. Unlike colorectal cancer, in which multiple primary cancers may be diagnosed, multiple diagnosed cancers of the prostate are extremely rare. Also, the prevalence of prostate cancer is higher than any other malignant cancers in men living in the United States. Therefore, we conducted a prevalence correction to the incidence rate of prostate cancer in the general population using the method described by Merrill and Morris (25). The prevalence correction excluded patients who already had prostate cancer from the general population because their relative frequency is not ignorable, especially in older men, due to the higher incidence rate and relatively good survival after diagnosis.

A standardized incidence ratio (SIR) was calculated as the ratio of observed to expected cases on the basis of attained age-, ethnicity-, and calendar year–specific standard rates. Therefore, SIRs are interpreted as the attained age-, ethnicity-, and calendar year–adjusted relative risks compared with the general population. Poisson regression models incorporating external standard rates were used to estimate SIRs and 95% confidence interval (95% CI; ref. 26). When the number of observed events was <20, we calculated exact 95% CIs based on Poisson distribution. We further calculated SIR separately for subgroups defined by age at first primary cancer diagnosis, duration of follow-up, and radiation therapy. As the use of prostate-specific antigen (PSA) test for prostate cancer screening began to increase since 1988 in the United States (27), we also divided the calendar period into the pre-PSA era (<1988) and PSA era (1988+) in the analysis of prostate cancer incidence to explore the potential influence of PSA test for prostate cancer screening.

Among the incident cases of prostate cancer developed after rectal cancer, we compared the lag time from rectal cancer to prostate cancer and tumor characteristics (stage, grade, PSA) using Wilcoxon rank-sum tests or χ² tests, as appropriate, between men treated for rectal cancer with radiation and men treated for rectal cancer without radiation. Overall survival after prostate cancer was also compared between the radiation therapy and nonradiation therapy groups using the Kaplan-Meier method and Cox proportional hazard model.

Table 1 shows the characteristics of the colorectal and prostate cancer cohorts. On average, members of all three cohorts were at their mid and late 60s when they had an initial diagnosis of cancer, with rectal cancer patients being ∼3 years younger than colon and prostate cancer patients. Of the 85,048 rectal cancer patients, 31.3% were treated with radiation therapy and virtually all of them were treated with external beam radiation. Notably, the proportion of rectal cancer patients receiving radiation therapy increased during the study period, starting from 13.7% in the 1970s to 37.6% during the period of 2000 to 2005. Few patients with colon cancer received radiation therapy. One third of prostate cancer patients were treated with radiation therapy, with external beam radiation being the most common form and brachytherapy being a minority. The proportion of prostate cancer patients receiving radiation therapy also increased during the study period, from 19.4% in the 1970s to 38.8% during the period of 2000 to 2005. The change in medical practice is more pronounced for brachytherapy, which increased from 0% in the 1970s to 10.0% during the period of 2000 to 2005.

After a total of 835,114 person-years of follow-up for the colon cancer cohort, 7,028 men developed subsequent prostate cancers, yielding an incidence rate of 841.6/100,000 person-years, which was similar to the general population (SIR, 1.03; 95% CI, 1.01-1.06; Table 2). However, the risk of prostate cancer was clinically significantly increased in patients who had colon cancer before age 50 years (SIR, 1.38; 95% CI, 1.18-1.60). After a total of 409,582 person-years of follow-up for the rectal cancer cohort, 2,680 men developed subsequent prostate cancers, yielding an incidence rate of 654.3/100,000 person-years, which was slightly lower than the general population (SIR, 0.88; 95% CI, 0.85-0.92). The risk of subsequent prostate cancer was decreased by 43% in rectal cancer patients who underwent radiation therapy (SIR, 0.57; 95% CI, 0.52-0.63) but not in rectal cancer patients who did not undergo radiation therapy (SIR, 0.99; 95% CI, 0.95-1.03) and in colon cancer patients who received radiation therapy (SIR, 0.97; 95% CI, 0.82-1.15). There were elevated risks for prostate cancer in both colon and rectal cohorts in the first 6 months of follow-up.

To explore the impact of radiation therapy for rectal cancer on subsequent occurrence of prostate cancer, we calculated SIRs separately by follow-up duration and calendar time (Table 3). After 6 months, the risk of prostate cancer was reduced in men with rectal cancer who received radiation therapy and the reduction persisted after 10 years of follow-up, suggesting that there is no latency effect. Interestingly, the apparent beneficial effect of radiation therapy only existed after year 1988.

We compared the characteristics of incident cases of prostate cancer between men with and without radiation therapy for rectal cancer (Table 4). The radiation therapy and nonradiation therapy groups have similar ages when they were diagnosed with rectal cancer but it took less time for men in the radiation therapy group to develop prostate cancer. The prostate cancers that developed in the radiation therapy group were at higher grade (poorly differentiated or undifferentiated) and later stage. They also had higher PSA levels than men in the nonradiation therapy group. Note that SEER collected tumor stage using different versions of American Joint Committee on Cancer's (AJCC) staging system in the study period as the AJCC continue to interpret and revise their staging system. Also, SEER started including PSA values in 2004. Nevertheless, survival analysis of these prostate cancer patients further suggests that men in the radiation therapy group had more aggressive form of prostate cancer than men in the nonradiation therapy group: The radiation therapy group has a median survival of 38 months compared with 70 months in the nonradiation therapy group.

After a total follow-up of 3,420,432 person-years in the prostate cancer cohort, 7,551 men developed colon cancers and 2,628 men developed rectal cancer. Overall, there was no association between prostate cancer and subsequent colorectal cancer (Table 5). In patients who had prostate cancer before age 50 years, the risk of subsequent colon cancer was marginally significantly increased (SIR, 1.51; 95% CI, 1.03-2.20). Overall, the risk of subsequent rectal cancer in prostate cancer patients who underwent radiation therapy was not different from the general population. However, after 10 years, there was a 44% increased risk of rectal cancer in these patients (SIR, 1.44; 95% CI, 1.22-1.71; Table 6). The relative risk of rectal cancer comparing prostate cancer with radiation therapy to those without radiation was 1.91 (95% CI, 1.52-1.89).

This study investigated a large cohort of patients diagnosed with colorectal cancer and prostate cancer within a national database and showed that the overall relationship between colorectal cancer and prostate cancer diagnoses was not clinically significant. However, several interesting findings emerged in the subgroup analyses. Patients diagnosed with colon cancer (but not rectal cancer) at younger ages are at moderately increased risk for subsequent prostate cancer. Patients diagnosed with prostate cancer at an early age were at an increased risk of developing subsequent colon cancer (but not rectal cancer). This two-directional association suggests that early-onset colon and prostate cancer may share similar genetic or environmental risk factors.

Consistent with previous studies of ours and others (22-24), we found that patients with rectal cancer undergoing radiation therapy had a 43% decreased risk of subsequent prostate cancer. Prostate cancer is one of the few malignancies that are not induced by irradiation exposure (28). There are several explanations for this risk reduction: (a) unrecognized confounders led to suspicious rate reduction, (b) radiation to the prostate tissue during adjuvant rectal therapy cures occult prostate cancer or prevents future neoplasia of the prostate, (c) radiation scattering to the scrotum induces hypogonadism and indirectly leads to a lower risk of prostate cancer, and (d) radiation therapy interferes with the screening of prostate cancer. We controlled for age, period, and ethnicity in the presented Poisson regression models and further controlled for tumor stage (data not shown) did not change the results. There was risk reduction in rectal cancer patients with radiation compared with both rectal cancer patients without radiation and the general population. Also, there was no risk reduction in colon cancer patients receiving radiation therapy, suggesting that the finding is less likely due to confounding. About the possible castration effect of radiation, one study showed that serum testosterone level only changed marginally in patients who received external beam radiation for prostate cancer. The author concluded that scattered testicular radiation played a less important role (29). In addition, there is no convincing epidemiologic evidence that androgens are strongly associated with prostate cancer risk (30).

We further evaluated the hypothesis that radiation therapy interferes with the screening of prostate cancer and found it as a likely explanation. Pelvic irradiation to the prostate either for prostate cancer or rectal cancer initially elevates serum PSA levels but may permanently impair PSA production (PSA levels were low for many years; refs. 31, 32). We showed that the rate of prostate cancer in patients receiving radiation therapy actually increased even higher than nonirradiated patients within the first 6 months, suggesting that the stronger surveillance effect may be due to the initial increase in PSA after radiation therapy. After 6 months, the prostate cancer incidence rate decreased in patients receiving radiation therapy and the reduction persisted after 10 years. The subgroup analysis by calendar time showed that the rate reduction only existed after 1988 when the use of PSA for prostate cancer screening became more commonly done in the United States. Two clinical trials in Sweden did not find that radiation therapy for rectal cancer decreased prostate cancer risk, possibly because PSA testing was not common in the study participants (33). Furthermore, we found that the prostate cancers that developed after radiation therapy for rectal cancers were at a later stage than those after rectal cancers without radiation therapy. These data support the hypothesis that radiation therapy interferes with prostate cancer screening because of its effect on PSA production.

We cannot completely rule out the possibility that the reduction in prostate cancer rates after radiation therapy is due to a direct biological effect on prostate. Because more poorly differentiated prostate tumors were observed after radiation therapy, it is possible that radiation therapy cures occult cancer or prevents subclinical cancer from progressing (as a result, higher-grade tumor is selected in). The observation of shorter latency from rectal cancer diagnosis to prostate cancer diagnosis in the radiation therapy group supports this hypothesis rather than the PSA screening hypothesis discussed above (in which one would expect delayed diagnosis).

There is a debate about the risk of a second primary rectal cancer after prostate radiation (5-14). Studies based on single institution or registry have had inconsistent results, which may contribute to a relatively small sample size. Interestingly, several studies based on the similar data from SEER cancer registries reached different conclusions (5-8, 11, 12), highlighting that appropriate analysis methods and interpretation are critical (34). Using SIRs, an earlier study and a most recent study found rectal cancer risk after prostate radiation was similar to the general population (8, 12). Using Poisson regressions and Cox proportional hazard models, two studies showed that patients who received radiation therapy for prostate cancer had an increased risk of subsequent rectal cancer compared with those who only had surgical therapy (5, 6) and concluded that the risk elevation is due to radiation therapy. Using Cox models, another study concluded that unmeasured confounders explained this association based on the finding that the relative risk of rectal cancer was 0.76 in patients receiving prostatic radiation and 0.34 in patients undergoing surgery compared with patients receiving neither treatment (7). Using logistic regressions, another study showed that radiation therapy group had 60% increased odds compared with nonradiation group (11), which was criticized for not fully adjusting for the duration of follow-up (34). We believe it is important to compare patients receiving radiation therapy with both patients not receiving radiation therapy and the general population. In the current study, we analyzed the same SEER databases with additional years of follow-up from 17 registries using Poisson regressions. We found that there was no excess risk of rectal cancer in patients receiving radiation therapy overall but the risk was 44% higher after 10 years of follow-up compared with the general population and 91% higher compared with patients not receiving radiation therapy. This observation is consistent with the theory that latency periods of ionizing radiation for most solid tumors are 10 years or longer.

There are several limitations in our study. Although the SEER database is a well-described database of high quality, some data are not collected. We controlled for age, race, and calendar time in the Poisson regression but other risk factors of rectal cancer or prostate cancer, such as diet and sedentary lifestyle, may be different between patients who received radiation therapy and those who did not. Dose of radiation therapy was not recorded in the database and radiotherapy treatment techniques are evolving in recent years. The elevated risk of rectal cancer 10 years after radiation therapy for prostate cancer is mainly applicable to beam radiation before 1995 because a few patients treated with brachytherapy were followed for >10 years. There seems to be a surveillance bias in the first 6 months after diagnosis of the primary cancer and less a concern later.

Our study has several strengths. To our knowledge, this is the largest study to investigate the association of prostate cancer and colorectal cancer as well as the impact of radiation therapy. Several additional years of follow-up and increased sample size facilitate subgroup analysis and help to answer several subtle research questions with certainty. We have carefully chosen our analysis strategy. In examining the relative risk of prostate cancer for rectal cancer patients receiving radiation therapy, we essentially used three comparison groups to help interpret the study findings: the general population, nonexposed rectal cancer patients, and colon cancer patients with radiation therapy. Prevalence correction was used in the calculation of prostate cancer incidence rates in the general population as we intended to estimate relative risk precisely. For example, the SIR of prostate cancer in the rectal cancer cohort was 0.96 (95% CI, 0.92-0.99) without prevalence correction but it changed to 0.88 (95% CI, 0.85-0.92) after prevalence correction. Stratifying the analysis into pre-PSA era and PSA era and examining characteristics of the second primary prostate cancer were helpful to elucidate the implication of the risk reduction in rectal cancer patients receiving radiation therapy.

In conclusion, we found that patients diagnosed with colon cancer (but not rectal cancer) under the age of 50 are at increased risk for subsequent prostate cancer and vice versa. Therefore, a diagnosis of colon cancer or prostate cancer in a younger aged man may require increased risk screening for prostate and colon cancer, respectively. We found that patients with rectal cancer who received radiation therapy are at decreased risk for subsequent prostate cancer. This may be due to the unintended benefit of radiation on a subclinical or premalignant prostate cancer or be related to interfering with the use of PSA as a tool for prostate cancer screening. If it is the latter, other modalities for prostate cancer screening or adjusting the threshold of PSA values will be necessary for this special population. Ten years after radiation therapy for prostate cancer, there is a mild increased risk of rectal cancer. This adverse event should be considered along with the benefits and other complications in the choice of treatments for prostate cancer patients.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Habibul Ahsan for his very helpful pieces of advice.