Journal of Cancer Research and Clinical Oncology

, Volume 140, Issue 6, pp 949–955

Multiple primary tumors following stage II and III rectal cancer in patients receiving radiotherapy, 1998–2010


    • School of Community Health SciencesUniversity of Nevada
  • George A. GoodwinIII
    • School of Community Health SciencesUniversity of Nevada
  • Jonathan Tay
    • Saint Mary’s Center for Cancer
    • Reno CyberKnife
Original Article – Cancer Research

DOI: 10.1007/s00432-014-1647-x

Cite this article as:
Smith-Gagen, J., Goodwin, G.A. & Tay, J. J Cancer Res Clin Oncol (2014) 140: 949. doi:10.1007/s00432-014-1647-x



This report investigated the impact of radiation therapy among stage II/III rectal cancer patients who were resected for cure and then developed second primary cancer.


The analysis included patients diagnosed with rectal cancer from 1992 to 2010 and who were registered in the National Cancer Institute’s Surveillance, Epidemiology and End Results database. Standardized incidence ratios assessed the location of second primary cancers by the receipt and sequence of radiation therapy. A Cox proportional hazards model examined the predictors for patients who developed second primary cancers.


The hazard ratio for developing any type of second primary was 12 % higher in patients receiving preoperative radiotherapy, Hazard Ratio and 95 % confidence interval, HR 95 % CI 1.12 (1.0, 1.2), and 33 % lower for patients receiving postoperative radiotherapy, HR 95 % CI 0.75 (0.7, 0.8), relative to patients who did not receive radiation therapy. The location of the second cancer varied by both the receipt and sequence of radiation therapy. Secondary rectal cancers were reduced 170 % after postoperative radiation and 103 % after preoperative radiation, compared to the non-receipt of radiation therapy. The impact of radiation therapy on secondary colon cancers was not as marked. Rectal cancer patients undergoing radiation therapy are at a higher risk of thyroid cancers and leukemia, but males have a lower risk of prostate cancer.


While preoperative radiation therapy is advantageous for reducing rectal cancer recurrence, this study identifies advantages of postoperative radiation for reducing second primary cancers. This research will help improve recommendations for postdiagnosis surveillance in patients with rectal cancer.


Rectal cancerSecond primary cancerRadiation therapyNational Cancer InstituteSEER


Over 40,000 patients are diagnosed with rectal cancer in the USA each year (American Cancer Society 2013), and the age adjusted rate in 2010 was 11.9 per 100,000 people (Howlader et al. 2013). Although the median age of a rectal cancer diagnosis is 69, the incidence rates of rectal cancer are increasing in patients younger than 40 years old. Between 1984 and 2005, the incidence rate for patients age 40 or younger increased 3.8 % per year (Meyer et al. 2010). Survival rates continue to increase, and many patients survive their rectal cancer diagnosis and treatment. The overall five-year relative survival rate for patients diagnosed between 2003–2009 was 67 % (Howlader et al. 2013). The survival rate was 88.2 % for local disease and 69.5 for regional disease (Howlader et al. 2013).

Cancer patients are surviving longer due to significant improvements in cancer treatment including radiation therapy, which reduces recurrence. However, concerns about late adverse events such as the development of second primary cancers are important issues for the increasing number of survivors (Ron 2003). Standard therapy for stage II–III rectal cancer includes surgery, chemotherapy and radiation treatment, although neoadjuvant radiation therapy is preferred. Rectal cancer and colorectal cancer need to be examined because standard therapy for colon cancer patients does not include radiation therapy (NCI 2013a, b; Sauer et al. 2004). Preoperative radiotherapy is generally preferred over postoperative radiotherapy because it reduces local recurrence and is anticipated that it may shrink the tumor enough to allow a sphincter sparing operation possible in a tumor that may otherwise require a permanent colostomy (Bujko et al. 2004; Hiotis et al. 2002; Bosset 2006). On the other hand, preoperative radiotherapy may result in overtreatment due to the less accurate clinical staging. This may be especially important in rectal cancer patients, first because overall survival is increasing and second because the incidence of rectal cancer in younger patients is also increasing. Younger patients receive radiation therapy more often than older patients (Fitzgerald et al. 2013).

The excess risk of second primary cancers for rectal and rectosigmoid junction cancers is between 7 and 13.5 per 10,000 person-years (Phipps et al. 2013). Second primary cancers may indicate the effects of treatment for the first cancer (Hemminki et al. 2001; Sawka et al. 2009; Koontz et al. 2013), or they may be caused by the same environmental or genetic factors that caused the first cancer (Dong and Hemminki 2001). Radiation therapy is associated with development of second primary cancers in several types of cancer. For example, thyroid cancer is one of the most common second primary cancers after radiotherapy for Hodgkin’s lymphoma during childhood. Secondary bladder cancers can occur after primary prostate cancers are treated with radiotherapy (Schonfeld et al. 2011; Abern et al. 2013). However, little is known about the risk of secondary tumors among rectal cancer survivors. The purpose of this study is to examine the impact of radiation therapy on outcomes in stage II/III rectal cancer patients who were resected for cure and who then developed a second primary cancer.


Study population

This report uses the Surveillance, Epidemiology and End Results (SEER) population-based cancer registry sponsored by the National Cancer Institute (SEER 2012). SEER is the only comprehensive source of population-based cancer information in the USA and is charged with reporting the incidence and mortality of the cancer burden in the USA. The registry collects demographic, clinical and treatment information from medical facilities. SEER also collects follow-up information on diagnoses of new cancer and in individuals with a history of cancer. The University of Nevada, Reno Office of Human Research Integrity exempted the study.

Patients in this study were diagnosed with a first primary rectal or rectosigmoid junction cancer between January 1, 1998, and December 31, 2010, according to the International Classification of Diseases for Oncology (ICD-0-2) codes C199 and C209. In this study, the researchers included patients with either stage II or stage III rectal cancer using the American Joint Committee on Cancer criteria (Edge et al. 2009) and those who had undergone curative surgical resection of the primary tumor (site-specific surgery codes 30, 40, 50, 60, 70 and 80). The researchers excluded patients diagnosed with first or subsequent cancers on a death certificate and only included patients with microscopically confirmed cancers. A total of 29,230 patients met the criteria.

Tumor definitions

SEER uses systematic methodology to define reporting rules for multiple primary tumors in order to reduce the confusion between secondary and multiple primary tumors. These rules include information regarding the site of the tumor, the behavior of the second tumor, the histology of all tumors, the date of the second diagnosis and laterality if it is a paired organ (Johnson et al. 2007). The general principles exclude tumors diagnosed within 2 months after the index rectal cancer, require that each primary cancer originates in a different site or tissue and exclude the possibility of relapse or metastasis from previous cancers. Only one tumor per organ is accepted unless there is a difference in morphology. Details can be found in SEER reporting rules (Johnson et al. 2007). A total of 2,424 patients who met the initial criteria developed a second primary cancer.


The primary variable of interest was the receipt of radiation therapy. SEER validation studies report radiotherapy as being 90 % complete (Cress et al. 2003). Radiation therapy was then grouped as no receipt of radiation therapy, any radiation prior to surgery or having received intraoperative or radiation after surgery. Next, the age at diagnosis, gender and racial groups were examined—Black, White, Asian and Pacific Islander. Finally, the impact of the index subsite (rectal or rectosigmoid junction), the stage at diagnosis (localized or regional) and a census-based income variable were examined. The income variable was used to adjust for socioeconomic status which may be indicative of unhealthy behaviors that put patients at risk for second primary cancers (Kreiger et al. 2002).


The distribution of the variables of interest was assessed using frequency counts and Kaplan–Meier median time to the development of a second primary cancer. The 95 % confidence intervals for the median survival times were calculated using a complementary log–log transformation (Lachin 2000).

The researchers calculated standardized incidence ratios (SIRs) by comparing the observed occurrence of second primary cancers after an index rectal or rectosigmoid junction cancer to the occurrence expected in the general population. The expected incidence rates were matched to the observed rates by gender, age in 5-year intervals, race (white, black and other) and calendar year. Thus, the SIRs represent the ratio of observed to expected secondary cancers adjusted for gender, age and calendar year. Confidence intervals were constructed using exact methods (19). This report considers the type of cancer that a patient developed after an initial diagnosis of rectal cancer stratified by receipt of radiation therapy.

Patients assessed in the time to second cancer analyses were followed until they were diagnosed with their second cancer, December 31, 2010, or death, whichever came first. Patients with unknown race and unknown sequence of radiation were excluded because they violate the proportionality assumption required for the Cox proportional hazards regression model. Cox proportional hazards regression was conducted to assess the impact of radiation therapy controlling for potential confounding variables and to examine whether radiation varied by the variables under consideration and effect modification with radiation therapy. SIRs were calculated using SEER*Stat version 8.1.2, and regression models were conducted using SAS version 9.1.


Distribution of the study population characteristics and SIRs are shown in Table 1. Overall, radiation after surgery decreased the incidence of a second primary cancer among rectal and rectosigmoid junction cancer patients, the SIR and 95 %  Confidence Interval (95 % CI) was 0.91 (0.8, 0.9). Non-receipt of radiation therapy was associated with a slight increase in a second primary relative to the general population, SIR and 95 % CI, 1.07 (1.02, 1.1). The location of the second primary cancer varied by the receipt and the sequence of radiation therapy. For primary rectal cancer patients who develop a secondary rectal cancer and who did not receive any radiation therapy, the risk was 220 % greater than in the general population, SIR and 95 % CI, 3.2 (2.7, 3.8) but among those patients who received radiation before surgery, the SIR was 117 % greater risk relative to the general population, SIR 95 % CI 2.17 (1.5, 3.0) and among these patients who received radiation after surgery the risk of a secondary rectal cancer was not statistically different than the general population, SIR and 95 % CI, 1.5 (0.9, 2.2). Thus, radiation prior to surgery represents a 103 % reduction in secondary rectal cancers (220 % − 117 % = 103 %), and radiation after surgery represents a 170 % reduction (220 % − 50 % = 170 %) relative to the general population. Although the risk of secondary colon cancers was elevated in patients with a primary rectal and rectosigmoid cancer, the impact of radiation therapy was not as clear. The largest effect was a reduction in prostate cancer after any type of radiation therapy, the SIR 95 % CI of 0.4 (0.3, 0.5), relative to those not receiving radiation therapy, SIR 95 % CI, 0.8 (0.7, 0.9). There were statistically significant increases in thyroid cancer after any receipt of radiation therapy. The SIR 95 % CI for radiation before surgery was 3.0 (1.7, 4.9), and SIR for radiation during or after surgery was 2.21 (1.2, 3.8). The incidence of urinary system and kidney cancers was increased after radiation prior to surgery, SIR and 95 % CI of 1.43 (1.1, 1.8) and SIR and 95 %CI for kidney cancers of 1.92 (1.3, 2.8). Statistically significant SIRs for secondary cancers with less than 48 patients per group were not included due to potentially inadequate power to detect statistically significant differences.
Table 1

Characteristics and SIRs of rectal and rectosigmoid junction cancer patients who develop second primary cancers by receipt of radiation therapy, SEER, 1998–2010


Incidence of developing a second primary cancer relative to the general population

No receipt of radiation therapy

Radiation prior to surgery

Radiation after surgerya

Overall (n = 2429)b

1.07 (1.02, 1.1)*

1.05 (0.96,1.1)

0.91 (0.8,0.9)*

Site of second primary cancerc


 Colon (n = 323)

1.64 (1.4, 1.9)*

2.22 (1.7, 2.8)*

1.4 (1.1, 1.9)*

 Rectum and anus (n = 191)

3.2 (2.7, 3.8)*

2.17 (1.5, 3.0)*

1.5 (0.9, 2.2)

 Prostate (males; n = 310)

0.8 (0.7, 0.9)*

0.40 (0.3, 0.5)*

0.38 (0.3, 0.5)*

 Urinary system (n = 265)

1.18 (1.0, 1.4)

1.43 (1.1, 1.8)*

1.16 (0.8, 1.5)

 Kidney (n = 88)

1.13 (0.8, 1.5)

1.92 (1.3, 2.8)*

1.48 (0.9, 2.3)

 Thyroid (n = 49)

1.54 (0.9, 2.4)

3.0 (1.7, 4.9)*

2.21 (1.2, 3.8)*

 Leukemia (n = 48)

0.68 (0.4, 0.9)*

0.99 (0.5, 1.7)

0.71 (0.3, 1.3)

Subsite of first primary cancer


 Rectum (n = 1,115)

1.1 (0.9, 1.1)

1.0 (0.9, 1.1)

0.89 (0.8, 0.9)*

 Rectosigmoid junction (n = 675)

1.10 (1.0, 1.2)*

1.34 (1.0, 1.8)*

0.96 (0.8, 1.1)



 Male (n = 1495)

1.0 (0.9, 1.1)

0.9 (0.8, 1.1)

0.77 (0.7, 0.9)*

 Female (n = 929)

1.12 (1.0, 1.2)*

1.36 (1.2, 1.6)*

1.2 (1.1, 1.4)*



 White and API (n = 2,238)

1.06 (1.0, 1.1)*

1.0 (0.9, 1.1)

0.9 (0.8, 0.9)*

 Black (n = 185)

1.30 (1.1, 1.6)*

1.2 (0.9, 1.6)

1.0 (0.7, 1.5)

Stage at diagnosis


 Local (n = 1,251)

1.1 (0.9, 1.1)

1.05 (0.9, 1.2)

0.92 (0.8, 1.1)

 Regional (n = 1,173)

1.12 (1.0, 1.22)*

1.05 (0.9, 1.2)

0.90 (0.8, 1.0)



<50 (n = 142)

1.66 (1.3, 2.1)*

1.90 (1.4, 2.5)*

1.40 (0.9, 2.0)

≥50 (n = 2,282)

1.06 (1.0, 1.1)*

1.0 (0.9, 1.1)

0.88 (0.8, 0.97)



 Low (n = 1,235)

1.08 (1.0, 1.2)*

1.08 (0.9, 1.2)

0.89 (0.8, 1.0)

 High (n = 1189)

1.07 (0.9, 1.2)

1.02 (0.9, 1.2)

0.93 (0.8, 1.1)

* Statistically significant differences compared to the general population

aIncludes interoperative surgery

bNumber of observed cases

cOnly subsites of second primary cancers with statistically significant SIR’s are shown

The hazard ratio for developing any type of secondary primary was about 12 % higher in patients receiving radiation before surgery and about 33 % lower for patients receiving radiation after surgery, all relative to the non-receipt of radiation therapy (see Table 2). This observation was not impacted by potentially confounding factors as documented by the similarity between the crude and adjusted hazard ratios. Put another way, the median time to a second primary cancer is significantly longer in patients who received radiation after surgery at 15.6 months, compared to those receiving radiation before surgery or no receipt of radiation, with median times of 14.1 and 14.5 months, respectively.
Table 2

Hazard ratios for risk of second cancers by statistically significant characteristics, SEER, 1998–2010


Median time (months)a

Risk (crude)

Risk (adjusted)

Receipt of Rx



14.5 (14.1, 14.8)



Before (vs. none)

14.1 (13.5, 14.8)

1.10 (1.0, 1.2)

1.12 (1.0, 1.2)

 After (vs. none)

15.6 (14.9,16.2)

0.79 (0.7, 0.9)

0.75 (0.7, 0.8)




14.8 (14.3, 15.1)




14.5 (14.0, 14.9)

1.10 (1.0, 1.3)

1.07 (0.9, 1.1)




14.6 (14.3, 15.0)




14.0 (13.6, 15.1)

1.23 (1.1, 1.4)

1.24 (1.1, 1.4)




14.7 (14.3, 15.1)




14.5 (14.2, 15.1)

1.12 (1.0, 1.2)

1.23 (1.1, 1.3)




16.2 (15.5, 17.0)




14.3 (14.1, 14.6)

2.05 (1.9, 2.5)

2.16 (1.9, 2.5)

aMedian time to second primary cancer (any type)

bAPI Asian and Pacific Islander were grouped with White because there were not significant differences between API and White. Sample sizes of other race groups were too small to assess

African Americans are 24 % more likely to be diagnosed with a second primary relative to Whites, Asian and Pacific Islanders (Table 2). This observation did not significantly change when secondary prostate cancers were excluded from the analysis (HR 1.22 (1.1, 1.4); data not shown in Table 2). Patients with regional stage cancers at diagnosis are more likely to develop second primary cancers regardless of receipt of radiation, and patients over age 50 are more likely to develop second primaries relative to younger patients regardless of receipt of radiation.

Stage of diagnosis of the index cancer modifies the effect of radiation therapy on the development of a second primary cancer, see Table 3. Specifically, radiation after surgery for regional rectal or rectosigmoid junction cancers reduced the hazard of a second primary cancer by 66 %. Radiation after surgery for localized rectal or rectosigmoid junction cancers reduced the hazard of a second primary cancer by 25 %. Radiation before surgery did not reduce the risk of secondary primary cancers. The interaction between sex and receipt of radiation therapy, although statistically significant, was not clinically different between the genders (data not shown).
Table 3

Significant effect modification by receipt of radiation therapy, surveillance, epidemiology and end results, 1998–2010

Stage at diagnosis

Hazard ratios (95 % confidence intervals)

No Rxt

Rxt before

Rxt after



0.9 (0.8, 1.1)

0.8 (0.7, 0.9)



0.92 (0.8, 1.1)

0.6 (0.5, 0.8)

RXT Radiation therapy


While preoperative radiation therapy is advantageous for reducing the recurrence of rectal cancers, this study identifies postoperative radiation as being the most advantageous for reducing second primary cancers. Patients with a primary rectal cancer are at greatest risk of secondary rectal, colon and thyroid cancers as well as leukemia, in descending order of risk. Secondary rectal cancers were reduced 170 % after radiation after surgery and 103 % after radiation prior to surgery, compared to the non-receipt of radiation therapy. For secondary colon cancers, the risk was greater if the patient received preoperative radiation and changed very little if the patient received postoperative radiation.

The results of this study are similar to Raj et al. who found that among patients with second primary locations in the colon and rectum, about 40 percent were in the rectum and 60 % were in the colon (Raj et al. 2011). This research extends their study question to examine locations beyond secondary colorectal cancers, examines the speed of the development of second primaries (using hazard ratios), controls for multiple covariates and focuses specifically on the receipt of radiation therapy.

Interestingly, the risk of a secondary prostate cancer in males was reduced after any first primary rectal cancer and this risk decreased even further with any type of radiation therapy (preoperative or postoperative). This observation is consistent with other research (Phipps et al. 2013). Potential reasons for the receded risks in prostate cancer are theorized to be the impact of health advice such as the reduction or elimination of smoking at the time of diagnosis. It is interesting to note that prostate cancer patients treated with radiation had a higher risk of cancer in areas that were irradiated, such as the rectum, but not in the remainder of the colon (Baxter et al. 2005). Research investigating the vulnerability of the prostate to radiation therapy might shed light on this issue as well. Furthermore, most US centers treat rectal cancers with a dose of 45–50 Gy in 25–28 fractions over 5–6 weeks, though some may have adopted the Swedish regimen of 25 Gy in 5 fractions over 1 week preoperatively. Radiobiologically, these 2 dose regimens may have a different impact on second malignancies. The Swedish and Uppsala trials showed a higher rate of prostate cancers (Birgisson et al. 2005) contrary to what this SEER data shows. Perhaps it is because the typical American dose given is therapeutic in eradicating or delaying very early, microscopic and undetected prostate cancer, whereas the Swedish regimen is not. This may be an area for future research.

In the descriptive analysis (SIRs), patients under age 50 were more likely to develop second primary cancers than older patients and more likely to develop secondary primary cancers following preoperative radiation than older patients. Inclusion of age groups less than 50 and over age 50 may have acted as a proxy for familial adenomatous polyposis (FAP) or hereditary non-polyposis colorectal cancer (HNPCC) since these patients are likely to develop their first primary cancers at age 50 or younger. These patients develop colorectal cancers earlier than other patients and have specific patterns of secondary primaries. However, patients with HNPCC are known to have a greater proportion of right-sided colon lesions and fewer cancers diagnosed in the rectum (Lichtenstein et al. 2000; Aarnio et al. 1999; Maul et al. 2006). These patients are also not known to develop a preponderance of cancers in the rectum (31). However, in the adjusted analysis, patients over age 50 had a shorter time developing a second primary cancer relative to younger patients. Adjustment for age did not change risk estimates regarding the sequence of radiation therapy, gender, race or stage of the second primary cancer.


The use of the SEER data has its strengths and limitations. In cohort studies, intense medical scrutiny after the first cancer may lead to over diagnoses, though we minimized this bias by only including microscopically confirmed cases. SEER lacks information on the patient’s receipt of chemotherapy, dosage of radiation therapy, smoking status, genetic information and recurrence.

Data on socioeconomic status (income) are present in the SEER database and may be a proxy to exposure to cigarette smoke, but the database does not include information regarding family history, FAP or HNPCC. The inclusion of age, acting as a proxy for FAP or HNPCC, most likely limited bias resulting from this missing information.

Furthermore, many secondary cancers develop decades after the primary cancer. This study only examined data between 1992 and 2010 and may have missed secondary cancers. A Swedish study examined data between 1952 and 1992 and found a significant misclassification of second primaries (Bergfeldt et al. 2000). However, diagnostic techniques have improved considerably over time and currently include microscopic confirmation through immunophenotyping and genetic studies. Furthermore, this research investigated not only the site of the secondary primary but risk factors for any type of second primary cancer. In this data, only 0.02 % of patients had second primary cancers in an unknown site.

In spite of these limitations, this study provides information on second primary cancers diagnosed in rectal cancer patient survivors that will serve as critical evidence to inform future research involving etiology and to help identify specific patient subgroups for whom more intense screening is warranted.


There are currently over one million rectal cancer survivors in the USA today (American Cancer Society 2013). Mortality from colorectal cancers has been declining 2.9 % per year over the past decade (33). These declines are no doubt due to preoperative radiotherapy which researchers have shown significantly reduces the local recurrence rate of rectal cancer (Swedish Rectal Cancer Trial 1997; Sauer et al. 2012; Benson et al. 2012). This research brings to light implications for second primary cancers as an end point.

The National Comprehensive Cancer Network recommends colorectal cancer survivors obtain a colonoscopy 1 year after surgical resection, repeat the examination in 3 years and then every 5 years if no advance adenomas are identified (National Comprehensive Cancer Network Practice Guidelines in Oncology 2011). However, there are no other guidelines for surveillance of other types of cancers. This research will help improve recommendations for postdiagnosis surveillance in patients with rectal cancer.


This study used the SEER database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the Applied Research Program, NCI; the Office of Research, Development and Information and the Surveillance, Epidemiology and End Results (SEER) Program tumor registries in the creation of this database.

Conflict of interest

We declare that we have no conflict of interest.

Copyright information

© Springer-Verlag Berlin Heidelberg 2014