Annual incidence of CCA
To assess the recent trends in CCA incidence, we identified all CCA cases from 2000 to 2016 in the SEER database. As shown in Figure 1A, the incidence of age-adjusted CCA was 3.3 per 100,000 persons in 2000, and increased to 8.8 per 100,000 persons in 2016; for comparison, the annual age-adjusted incidence of all malignancies is also depicted.
In addition, we divided all CCA cases into different subgroups based on the age and race of the patient as well as grade and stage of the tumor. First, the age-specific incidence was calculated for three age groups: <60 years, 60–69 years, and >70 years. As shown in Figure 1B, the incidence of CCA increased dramatically from 2000 to 2016 in patients aged <60 years, with nearly a 3-fold rise to 4.7 per 100,000 persons; among those aged 60–69 years or >70 years, there was a more modest increase of 2.5-fold.
Among the tumor grade groups, the most dramatic rise in incidence was described in patients with grade IV CCA (from 0.1 per 100,000 persons in 2000, to 0.7 per 100,000 persons in 2016; Figure 1C). Among the tumor stage groups, the incidence of localized CCA increased the most relative to regional or distant (from 2 per 100,000 persons in 2000 to 6.2 per 100,000 persons in 2016; Figure 1D). Among the ethnic groups, the incidence of CCA increased the most in Caucasians (from 2.1 per 100,000 persons in 2000 to 8.4 per 100,000 persons in 2016; Figure 1E).
Overall, according to the SEER 18 data, the incidence of CCA cases diagnosed increased statistically significantly from 2000 to 2016 (annual percent change (APC): 4.6, 95% confidence interval (CI) [3.4, 5.7], P < 0.05).
Patient characteristics
To compare the demographic and clinical characteristics of CCA at the different primary tumor sites, we analyzed 104,206 CCA patients identified in the SEER database (Table 1). We found that the median age of these patients at diagnosis was 62 years (range: 53–71 y), and the median OS was 46 months (range: 17–91 mo).
Comparing the different primary tumor sites, the lowest median age at diagnosis (56 y) was that of patients with ovarian CCA, and the highest (71 years) was for patients with bladder CCA. The median OS was best for CCA of the prostate (79 mo) and breast (72 mo), while the median OS was worst for CCA of the pancreas (3 mo) and liver (6.5 mo). All of these differences in OS were significant (P < 0.001; Supplementary Figure 1A).
The primary tumor site in these patients was significantly associated with the age at diagnosis (P < 0.001; Table 1). Patients aged ≥61 years were more likely to develop primary CCA of the KRP, corpus and cervix uteri, lungs, bladder, pancreas, prostate, or liver; while patients aged 31-60 years were more likely to have CCA of the POM. Moreover, the tumor grade was significantly different between the various primary tumor sites (P < 0.001; Table 1): patients with grade II tumors were more likely to have CCA of the KRP; while patients with grade III tumors were more likely to have CCA of the corpus and cervix uteri, lungs, breast, or prostate.
The tumor stage was significantly different between the various primary tumor sites (P = 0.001; Table 1): localized CCA was more likely to be in the KRP, ovary, cervix uteri, breast, prostate, or liver; while distant CCA was more likely to be in the lungs or pancreas. In addition, we determined that the metastatic sites differed depending on the type of CCA. Patients with CCA of the KRP, cervix uteri, or vagina were most likely to have lung metastases; patients with CCA of the ovary, bladder, or pancreas were more prone to have liver metastases; and patients with CCA of the lung, breast, or prostate were tended to have bone metastases (Supplementary Figure 1B).
Survival
We identified the latest trends in the survival of all CCA cases from 2000 to 2011 in the SEER database, relative to the general population. As shown in Figure 2A, the age-standardized 3-year and 5-year relative survivals increased from 2000 to 2011, rising by 9.1% and 9.5%, respectively. Specifically, when we examined the CCA cases by grade (Figure 2B–C), we found that the age-standardized 3-year and 5-year relative survivals of patients with grade I CCA increased from 89.6% and 84.0% in 2000 to 98.1% and 96.0% in 2011, respectively. Meanwhile, the age-standardized 3-year and 5-year relative survivals of patients with grade II CCA slightly improved from 85.4% and 79.7% in 2000 to 93.6% and 90.0% in 2011, respectively.
Those with grade III–IV CCA showed an even greater improvement: the 3-year and 5-year relative survivals of grade III patients increased from 66.7% and 41.6% in 2000 to 79.8% and 63.5% in 2011, respectively. The 3-year and 5-year relative survivals of the grade IV patients increased from 60.9% and 39.6% in 2000 to 73.8% and 51.1% in 2011. When we examined the age-standardized 3-year and 5-year relative survivals by tumor stage, we found that the relative survival of patients with localized, regional, or distant tumors had improved slightly over time (Figure 2D–E).
The age-standardized 3-year and 5-year survivals of the CCA patients relative to the general population and according to the primary tumor site were analyzed (Figure 3A). The largest change in 3-year to 5-year survival was for CCA of the POM (46.8% to 32.2%) or prostate (72.3% to 59.2%), and the smallest change was for CCA of the pancreas (11.4% to 8.3%) or KRP (85.6% to 81.1%). We examined the known 3-year and 5-year relative survivals of CCA of different primary tumor sites according to the tumor stage (Figure 3B–C). The best 5-year relative survival for regional and distant tumors was for patients with ovarian CCA.
We performed a multivariate analysis and calculated the hazard ratio for OS (Table 2). Age, gender, year of diagnosis, marital status, ethnicity, grade, stage, and primary tumor site were all significantly associated with OS. We found that women (HR, 0.85; 95% CI, 0.83–0.87) had a better OS than men, and patients with grade III (HR, 1.21; 95% CI, 1.16–1.26) or grade IV (HR, 1.68; 95% CI, 1.59–1.77) CCA had a worse OS than did those with grade I CCA. However, the OS was not statistically different between grade II and grade I CCA. After adjusting for other variables, regional CCA (HR, 1.91; 95% CI, 1.86–1.97) and distant CCA (HR, 8.29; 95% CI, 8.06-8.53) had a worse OS than did localized CCA. Compared with CCA of the KRP, patients with CCA of the liver had the worst OS (HR, 4.89; 95% CI, 3.71–6.44), those with CCA of the pancreas had the second worst OS (HR, 3.70; 95% CI, 2.99–4.59), and those with CCA of the ovary had the best OS (HR, 0.75; 95% CI, 0.71–0.79).
We analyzed the latest trends of the OS during 2000–2007 and 2008–2016. Compared with 2000–2007, the risk of death in CCA diagnosed in 2008–2016 was less by 13% (HR, 0.87; 95% CI, 0.85–0.89). We calculated the 3-year and 5-year hazard ratios through multivariate analysis, and the patients with grade II CCA had better 3-year (HR, 0.84; 95% CI, 0.79–0.90) and 5-year (HR, 0.88; 95% CI, 0.83–0.93) survivals than did those patients with grade I CCA. All of the above comparisons are significant (P < 0.001).