Second primary colorectal cancer after the initial primary colorectal cancer

This study used a general dataset that selected data from a public health program and therefore did not require institutional review board approval. We identified patients with diagnosed CRC reported by the Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute between 1992 and 2012.

All cases of IPCRC were: histologically confirmed colorectal cancer and pathological type is adenocarcinoma or mucinous adenocarcinoma [18]. Cases where diagnosis of IPCRC was confirmed by death certificate only or autopsy only were excluded. A minimum interval of 6 months between the diagnoses was required to exclude synchronous cancers (latency period).

Three sites were evaluated: the right colon (cecum, ascending colon, hepatic flexure, and transverse colon); the left colon (splenic flexure, descending colon, and sigmoid colon), and the rectosigmoid colon (rectosigmoid junction and rectum). The SEER stage included localized (malignant but confined to primary organ), regional (invasion beyond primary organ but no distant metastasis), or distant (with distant metastasis) and unknown [19]. The study cohort was divided into subgroups based on latency intervals: 0–4 years, 4–8 years, > 8 years. Individuals within each cohort (e.g. LCC, RCC, rectosigmoid cancer) were classified by whether they had an ensuing diagnosis of second primary malignant neoplasm of CRC, hereafter referred to as SPCRC.

Survival time was defined as the number of months from study commencement until the date of death from any cause, or until the end of the observation period. Patients alive at the end of follow-up were censored.

Standardized incidence ratios (SIRs) and absolute excess risk (AER) values were obtained from the SEER program using SEER*Stat software. SIRs were computed by comparing the observed (O) number of second malignancies with those expected (E) in the general population to quantify second malignancy risks to assess excess relative risk of SPCRC compared with the general population. Within SEER, the AER is defined as the excess cancers per 10,000 persons per year, which is calculated as follows: ([observed count-expected count] × 10,000)/ person-years at risk [20]. Joinpoint regression was used to analyze SPCRC incidence and mortality risk trends. We use annual percent change (APC) to assess rate changes.

Multivariable logistic regression was used to determine which prognostic factor were associated with the presence of SPCRC at diagnosis. The Kaplan-Meier method was used to compare survival outcomes.

A P value of 0.05 was considered statistically significant and P values were two sided. Analyses were performed with the SAS version 9.2 software package (SAS Institute, Cary, North Carolina) and the survival package within the R version 2.11 statistical software package (http://​www.​r-project.​org) was used.

After exclusions, together, among all the IPCRCs, 2.0% (4977/252,404) were considered to have 2nd CRC, and the incidence in those with RCC was 2.3% (1756/96,754), 2.3% (1770/73,132) in those with LCC, and 1.8% in those with rectal cancer (1451/77,550) (Additional file 1: Table S1). Those aged between 70 and 79 years had the highest proportion (31.8%) of SPCRC and this was the same for other CRCs (35.4% for RCC, 31.2% for LCC, and 28.3% for rectosigmoid cancer). Based on the SEER staging system, 1989 (40.0%) SPCRCs were considered to be localized disease, 1703 (34.3%) had regional disease, 888 (17.8%) had metastatic lesions, and the remainder (8.0%) of the patients unstated. When we analyzed the 2nd CRC distribution by histology grade, we found that moderately differentiated cancer accounted for the largest proportion of SPCRC, which was apparently higher than any of the other histology groups. More than half of 2nd CRCs were diagnosed in the first four years in but approximately 20% of 2nd CRCs were identified in the following 4–8 years after the 1st CRC. Patients with surgery history (95.1%) were more likely to develop SPCRC compared to patients without surgery history (4.9%). For patients without surgery, they are more likely to be in the early stage and had better prognosis, then the chance to have the SPCRC is bigger. However, for patients without surgery history, they may tend to be in the advanced stage and the prognosis is poor, so the probability to have the SPCRC is relatively reduced.

Table 1 compares the demographic and clinical features of second primary CRC between RCC, LCC, and rectal cancer. RCC accounted for nearly half of these SPCRCs (2144/4977). Patients tend to have more localized disease, be of white ethnicity and have moderately differentiated disease at diagnosis of the second primary RCC. The median time to the 2nd RCC, LCC, and rectosigmoid cancer was 38.0 months, 28.0 months, and 31.0 months, respectively. The mean age exceeded 70 years of age and the eldest was in the RCC cohort.

Figure 1 shows the trends in incidence rate of CRC in the standard US population and the trend in the excess risk of a second primary CRC following latency time. SPCRC incidence has declined since the first year of the diagnosis of the IPCRC (Fig. 1a). Compared with the general population, the risk summit for SPCRC appears in the second year of the IPCRC diagnosis in the study cohort (Fig. 1b). In the first 8-years during follow-up, the risk of second CRC cancer patients was significantly higher compared with the general population and then tended to have similar likelihood as that of the general population with intervals of > 8 years for diagnosis (Fig. 1b). Similar results were obtained focusing on LCC, RCC, and primary rectal cancers (Additional file 1: Figure S1).

Overall CRC risk was significantly higher than expected in the general population after CRC (SIR, 1.73; 95% CI, 1.69–1.78), RCC (SIR, 1.52; 95% CI 1.45–1.59), LCC (SIR, 1.98; 95% CI, 1.89–2.08), and rectosigmoid cancer (SIR, 1.77; 95% CI, 1.67–1.86) (Additional file 1: Table S2). The calculated excess risk (beyond the expected number) was 15.38 per 100,000 in the whole cohort, 12.09 in the RCC, 20.33 for the LCC, and 14.21 for rectosigmoid cancer. The resulting SIR data and AER values are given in Additional file 1: Table S2. Furthermore, the 2nd CRC malignancies risk was elevated for LCC (SIR, 1.73; 95% CI, 1.64–1.83), RCC (SIR, 1.65; 95% CI, 1.56–1.72), and rectosigmoid cancer (SIR, 2.04; 95% CI, 1.94–2.15). Stratified analyses suggested that the risk might be a consistent phenomenon across almost all other subtypes. Of note, patients diagnosed at younger ages (< 40 years) were particularly more prone to 2nd CRC (SIR, 14.78; 95% CI, 12.20–17.74 for the 2nd CRC; SIR, 15.15; 95% CI, 10.49–21.17 for the 2nd RCC; SIR, 10.42; 95% CI, 6.67–15.50 for the 2nd LCC; SIR, 17.88; 95% CI, 13.54–23.16 for the 2nd rectosigmoid cancer).

As is shown by the multivariate Cox regression in the Table 2, History of LCC (LCC vs RCC, HR = 1.369, 95% CI, 1.269–1.477, P < 0.001), History of rectosigmoid colon cancer (rectosigmoid colon cancer vs RCC, HR = 1.090, 95% CI, 1.003–1.184), age > 60 (60–69 vs < 40, HR = 1.342, 95% CI, 1.073–1.679, P = 0.010; 70–79, HR = 1.541, 95% CI, 1.234–1.925, P < 0.001; > 80 vs < 40, HR = 1.339, 95% CI, 1.066–1.680, P = 0.012), Female (male vs female, HR = 0,930, 95% CI, 0.871–0.992, P-0.028), Black (White vs Black, HR = 0.490. 95% CI, 0.447–0.536, P < 0.001; Hispanic/Latino, HR = 0.604, 95% CI, 0.528–0.690, P < 0.001; Asian and Pacific, HR = 0.798, 95% CI, 0.710–0.898, P < 0.001), without surgery history (No vs yes, HR = 1.657, 95% CI, 1.383–1.985, P < 0.001), well differentiated pathology (moderately vs well, HR = 0.868, 95% CI, 0.785–0.961, P = 0.006; poorly vs well, HR = 0.814, 95% CI, 0.720–0.919, P = 0.001; undifferentiated vs ell, HR = 0.524, 95% CI, 0.354–0.775, P = 0.001) were the risk factors of the presence of SPCRC. Figure 2 shows the relative importance of each independent variable and shows that race in ISPCRC diagnosis was the most important factor in predicting SPCRC risk, followed by age at diagnosis and primary tumor site. Compared with a reference race of black ethnicity, survivors of other race had substantially reduced risks of SPCRC, with a hazard ratio (HRs) < 0.799. Compared with those aged < 40 years, the age group 70–79 had the highest risk of developing SPCRC (HR = 1.516, P < 0.001). Those with LCC had an increased risk of SPCRC compared with a reference RCC (HR = 1.362, P < 0.001) and the trend still remained after accounting for other confounding factors.

We used follow-up data to compute actual observed survival estimates comparing patients with SPCRC and those with IPCRC. The estimates are shown in Fig. 3. The survival estimates for the IPCRC and the SPCRC groups were different (log-rank, P < 0.001). We generated separate OS comparison data for categorized sites. For RCC, LCC, and rectosigmoid cancer, those with SPCRC had an apparent poorer survival outcome than the IPCRC groups (P < 0.001 for RCC, LCC, and rectosigmoid cancer, respectively, Fig. 3). We then further compared the OS between those with SPCRC by cancer sites and we found that those with RCC had the best survival outcome (P < 0.001for the whole cohort; P = 0.002 for RCC vs LCC; P < 0.001 for RCC vs. rectosigmoid cancer; and P = 0.352 for LCC and rectosigmoid cancer) (Fig. 4).