Splenectomy and increased subsequent cancer
risk: a nationwide population-based cohort study
Li-Min Sun, M.D.a, Hsuan-Ju Chen, M.Sc.b,c, Long-Bin Jeng, M.D.d,e,
Tsai-Chung Li, Ph.D.f,g, Shih-Chi Wu, M.D.e,h, Chia-Hung Kao, M.D.e,i,*
The spleen is the largest lymphatic organ in the human body and serves a crucial role in immune function. It combines the innate and adaptive immune system, and is considered to be the most vital organ for antibacterial and antifungal immune reactivity.1,2 Splenectomy can be performed both as an emergency procedure following abdominal
trauma and electively when indicated for
hematological disorders.3,4 In addition to infection, cancer is thought to be related to human immunity.5,6 Splenectomy has been studied for its effects on tumor cell growth with inconsistent results,7–10 and controversial animal studies have revealed the influence on the antitumor immune system after splenectomy in vivo.11,12 Few studies have explored the clinical human cancer risk after splenectomy; however, epidemiologic studies have observed increased risks among certain solid tumors and hematological malignancies.13–15 Malignant neoplasm has been the leading cause of death for the Taiwanese general population since 1982, and is becoming a crucial public health problem with aging
populations and changes in lifestyles.16 The Taiwan government has focused on this concern and launched primary
and secondary prevention programs for major cancers, such as high-risk factor avoidance and periodic cancer screening to reduce cancer incidence and mortality rates.16 Therefore,
population-based investigations regarding cancerpreventive epidemiology continue to be conducted. Based
on the limited available clinical reports highlighting the association between splenectomy and subsequent cancer risk,
we hypothesized that people with a splenectomy in Taiwan may be at increased risk of cancer. Our results were generated from a retrospective cohort study for people with a
splenectomy. The original database was derived from the National Health Insurance (NHI) system in Taiwan.
Methods Data source
The NHI program was initiated in Taiwan in 1995 and currently covers approximately 99% of the 23 million population of Taiwan. In this study, we established a longitudinal cohort study from the Taiwan National Health Institute Research Database (NHIRD), with hospital admissions of the entire population insured in Taiwan between
1996 and 2011, which contains the healthcare claims of each patient. To protect privacy, the NHIRD scrambled data cryptographically on patient identities and institutions.
Each person in Taiwan has a unique personal identification number, and all NHI datasets can be interlinked with the personal identification number of each person. For this study, disease diagnosis was based on disease records according to the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) in
inpatients and catastrophic registry files. This study was approved by the Ethics Review Board of China Medical University (CMU-REC-101-012).
Study population
Initially, we identified 10,964 patients with newly received splenectomy (ICD-9-CM code 41.5) from 2000 to 2006 from the inpatient registry file. We excluded patients with less than 20 years of age (n 5 1,067), diagnosis of cancer at baseline (n 5 1,280), and less than 1-year of follow-up (n 5 3,719) from the analysis. The
remaining 4,898 patients with newly received splenectomies were included in the splenectomy cohort, which was
divided into 2 groups based on trauma status. Four study subjects in the no splenectomy cohort for every patient with a splenectomy were randomly selected from insured people without a history of splenectomy or cancer, and frequency matched with age (per 5 years), sex, diabetes, hypertension, hyperlipidemia, coronary artery disease, and index year.
Overall, 4,898 patients were included as the splenectomy cohort and 19,592 patients as the no splenectomy cohort.
The sociodemographic status included sex and age (20 to 44 and R45 years). Information on pre-existing comorbidities, including diabetes (ICD-9-CM code 250.xx), hypertension (ICD-9-CM code 401.xx to 405.xx), hyperlipidemia (ICD-9-CM code 272.xx), and coronary artery disease (ICD-9-CM 410.xx to 414.xx), were also recorded.
The primary outcome was cancer occurrence (ICD-9- CM code 140.xx to 208.xx), which was determined by linking records with the catastrophic illness registry of the NHIRD. All subjects were followed until patient withdrawal from the NHI program, cancer development, death,
or the end of 2011. We attempted to minimize the impact of any effect of existing medical conditions on splenectomy by excluding cancer that occurred in subjects during the first year of follow-up. We also investigated the risk of individual cancer sites between the splenectomy cohort and the no splenectomy cohort. The cancer sites were separated into head and neck cancers (ICD-9-CM code 147.xx for nasopharynx cancer; ICD-9-CM code 140.xx to 146.xx and 148.xx to 149.xx for other head and neck cancers),
esophagus cancer (ICD-9-CM code 150.xx), stomach cancer (ICD-9-CM code 151.xx), colorectal cancer (ICD-9-
CM code 153.xx and 154.xx), liver cancer (ICD-9-CM code 155.xx), lung cancer (ICD-9-CM code 162.xx), breast cancer (ICD-9-CM code 174.xx), gynecologic cancer (ICD-9-CM code 179.xx to 184.xx), prostate cancer (ICD-9-CM code 185.xx), bladder cancer (ICD-9-CM code 188.xx), non-Hodgkin’s lymphoma (ICD-9-CM code 200.xx and 202.xx), Hodgkin’s disease (ICD-9-CM code
201.xx), leukemia (ICD-9-CM code 204.xx to 208.xx), and other cancers.
Statistical analysis
We compared the distribution of sociodemographic
factors and pre-existing comorbidities between the cohorts
with and without a splenectomy using the chi-square test for category variables and the t test for continuous variables.
For estimating the cumulative incidence of cancer risk in the splenectomy cohort and the no splenectomy cohort, we performed survival analysis using the Kaplan–
Meier method, with significance based on the log-rank test. The Cox proportional hazards model was used to assess the risk of the study population developing cancer
and to produce hazard ratios (HRs) and 95% confidence intervals (95% CIs). Statistical analyses and data management
were performed with SAS 9.3 (SAS Institute, Inc,
Cary, NC), and the cumulative incidence curves were determined using R computer software (R Foundation for Statistical
Computing, Vienna, Austria). The significance levels were set at a 2 tailed P value of .05.
Results
We established 4,898 splenectomy patients and 19,592 subjects without splenectomy, with the same average age (mean 5 45 years) and more men than women (62.03% vs 37.97%) (Table 1). The distributions of age, sex, and comorbidities were the same in both groups. The mean follow-up periods for the splenectomy and nonsplenectomy groups were 7.41 (62.72) and 7.90 (62.52) years,
respectively.
The cumulative subsequent cancer incidence curves showed that the curve for the splenectomy cohort was significantly higher than that for the no splenectomy cohort (P , .001 for the log-rank test; Fig. 1). Patients with splenectomy were 1.94 times more likely than the no splenectomy
cohort to experience cancer (9.73 vs 5.03 per 1,000 person-years), with an adjusted HR of 2.06 (95% CI 1.81 to 2.33) after adjusting for age, sex, and history of diabetes, hypertension, hyperlipidemia, and coronary artery disease
(Table 2). Among patients receiving splenectomy, patients without and with trauma exhibited a significantly increased risk of overall cancer compared with the no splenectomy cohort (HR 2.64, 95% CI 2.30 to 3.05 and HR 1.29, 95%
CI 1.05 to 1.60, respectively). Sex-specific analysis showed that women exhibited a 2.16-fold higher risk of developing cancer (95% CI 1.76 to 2.65) and men exhibited a 1.99-fold (95% CI 1.70 to 2.34) higher risk of developing cancer than the no splenectomy cohort. After age stratification, we observed that both the age groups of 20 to 44 years and
45 years and older of splenectomy patients exhibited a significantly higher risk of cancer than the no splenectomy
cohort. Regardless of subjects without or with comorbidities, the risk of cancer was higher in the splenectomy
cohort than that in the nonsplenectomy cohort.
We observed a significantly positive association between splenectomy and the risks of site-specific cancer, including other head and neck, esophagus, stomach, liver, non- Hodgkin’s lymphoma, and leukemia (Table 3). Splenectomy patients without trauma were at significantly
increased cancer risk of head and neck, other head and neck, esophagus, stomach, liver, breast, non-Hodgkin’s lymphoma, and leukemia. On the other hand, splenectomy patients with trauma were only at significantly increased cancer risk of esophagus, stomach, liver, and leukemia.
We calculated the HR for specific cancer types according to splenectomy status stratified by follow-up years, and the results are shown in Table 4. For overall cancer, higher risks for developing cancer were observed in the follow-up periods of 1 to 3, 4 to 6, and 7 years or more. In addition, the risks for developing cancer were observed in the followup periods of 1 to 3, 4 to 6, and 7 years or more for splenectomy patients without trauma, as well as in the
follow-up periods of 7 years or more for splenectomy patients with trauma. For head and neck cancer, a 3.99-fold
risk of developing cancer was found in the 7 year or more follow-up period. For other head and neck cancer, a 3.83-fold risk of developing cancer was found in the 7
year or more follow-up period. For esophagus cancer, a 4.54-fold risk of developing cancer was found in the follow-up period of 4 to 6 years. For stomach cancer, the risks for developing cancer were observed in the followup periods of 1 to 3 years and 7 years or more. For liver cancer, higher risks were observed for developing cancer in the follow-up periods of 1 to 3, 4 to 6, and 7 years or more. For non-Hodgkin’s lymphoma, a 15.23-fold risk of developing cancer was found in the follow-up period of 4 to 6 years. For leukemia, higher risks for developing cancer were observed in the follow-up periods of 1 to 3 years.
Comments
This population-based cohort study revealed that Taiwanese with splenectomy had a significantly higher risk of
subsequent overall cancer, as well as certain gastrointestinal tract cancers, other head and neck cancers, and
hematological malignancies. The phenomenon was more
obvious for patients with nontraumatic reasons for splenectomy.
Further stratified analysis by follow-up years
revealed various patterns among diverse cancers, which are likely caused by the coincidence effect of splenectomy and cancer.
The age-adjusted incidence rate of cancer in Taiwan has increased steadily, and overall cancer incidence rates
increased by 2% per year between 2008 and 2011, reaching 295.1 (invasive cancers only) and 320.7 new cases (carcinoma in situ and invasive cancers) per 100,000 persons in
2011.17 Despite enhanced detection, much of the increased incidence may be associated with aging populations and a
westernized lifestyle.16 In contrast, trend analysis from surveillance epidemiology and end results data showed that
the overall cancer incidence rates for all racial and ethnic groups combined in the United States decreased by .6%
per year between 2006 and 2010.18 This obstacle continues to be a challenge for public health in Taiwan, and has aroused the attention of the government to enact policies
regarding cancer prevention. Prevention is the most costeffective long-term control strategy for cancer, with an estimated
one third of all cases considered avoidable.19 One of
the research topics in this field is to explore the possible risk/protective factors of cancer. The NHI program in
Taiwan is a suitable resource for providing valuable materials with which to approach populated-based studies. Our
research team recently performed a cohort study by using the NHIRD and found that Taiwanese patients with diabetes are at an elevated risk of cancer (particularly
colorectal and pancreatic cancers), and the use of thiazolidinediones might decrease the risk of liver cancer in diabetic
patients.20 This information inspires us to conduct
studies to discover other factors that may be related to cancer.
Infection is a well-known complication of splenectomy,
21–23 and immunocompromised patients are also
vulnerable to certain types of cancer. Therefore, it is plausible to hypothesize that people with splenectomy may
have a higher risk of developing cancer.
Besides age, sex, and index year, our control subjects were frequency matched to cases with comorbidity as well.
Diabetes, hypertension, and hyperlipidemia are suggested to be a risk factor for certain types of cancer.22,24–29 In addition, we also adjusted these comorbidities and other factors in the multivariate analysis to eliminate possible bias. Table 2 shows that splenectomized patients had a significantly higher adjusted HR of overall cancer compared with the no splenectomy participants, and
Table 4 shows that the significantly higher risks in the splenectomy group were consistent across various follow-up
years. This finding is consistent with a cohort study from American veterans, which showed that the cancer risk in splenectomy patients was significantly higher across various follow-up years [relative risks (95% CIs) are 2.05 (1.82 to 2.32), 1.44 (1.28 to 1.62), and 1.35 (1.24 to 1.47)
for 2 to 5, 5 to 10, and greater than 10 years latency, respectively].
13 Table 2 also highlights the higher risks of overall
cancer in the splenectomy cohort, regardless of sex, age, and comorbidity. When divided splenectomy patients by trauma and nontrauma reasons, we found that both groups
had a significantly higher risk of overall cancer compared with nonsplenectomy cohort, but the same pattern regarding consistently higher risk across sex, age, and comorbidity only limited to nontraumatic reasons.
For specific cancer site analysis, we separated nasopharyngeal carcinoma from head and neck cancer because
nasopharyngeal carcinoma is relatively common in Chinese people.30 Splenectomy patients were found to have significantly higher risks in other head and neck, esophagus, stomach, liver, non-Hodgkin’s lymphoma, and leukemia cancers
(Table 3). Kristinsson et al13 conducted a cohort study to evaluate long-term risks after splenectomy among 8,149 cancerfree American veterans and found significantly higher risks of being hospitalized for all cancer, as well as for esophagus, liver, pancreas, lung, prostate, non-Hodgkin’s lymphoma, and leukemia cancers among splenectomy patients. Their further sensitivity analyses for subset of patients with diagnoses of autoimmune disease or trauma did not change the overall results.13 Although both trauma and nontrauma reasons had higher risks for developing some kinds of cancer in our splenectomy patients, Table 3 indicates that a significantly
higher risk more likely occurred among those with nontraumatic reasons. The exact mechanism for the possible
association between splenectomy and cancer risk remains unclear. A plausible mechanism is that the spleen is thought to be involved in immunologic defenses, and provides active immune response through humoral and cell-mediated pathways.
Splenectomy may impair immune surveillance in the
host.21,31 Cancer is suggested to be related to human immunity,
5,6 and immunity may therefore partially account for the possible linkage between splenectomy and cancer risk. Our data show a significantly increased risk of gastric cancer in the splenectomy group. We conjecture that splenectomy patients are susceptible to bacterial infection. Infection with
Helicobacter pylori causes chronic inflammation and is the strongest risk factor for the development of gastric cancer.32 Our data identified that a significantly higher risk of non- Hodgkin’s lymphoma was only observed in the nontraumatic
group, but both the traumatic and the nontraumatic groups had a significantly higher risk for leukemia. A previous population-based report from Denmark was conducted to evaluate cancer risk after splenectomy for both traumatic and nontraumatic reasons. The authors found no increased risk for cancer among patients who underwent splenectomy because of trauma. However, they indicated an increased risk for certain hematological malignancies in patients who underwent splenectomy for nontraumatic reasons.15 Another epidemiologic study with a long-term follow-up of 740
American veterans splenectomized because of trauma during WorldWar II showed no increased risk of cancer.33 This null
association between total or site-specific cancers and splenectomy for external trauma reason has also been reported
in Swedish residents.14 On the other hand, Kristinsson
et al13 performed subanalyses restricted to splenectomy patients for trauma and observed significantly elevated risks
of cancer in splenectomy patients, which is consistent with our findings. Spleen is thought to be involved in immunologic defenses and splenectomy (no matter trauma or nontrauma reasons) may be possibly linked to the tumorigenesis. For nontraumatic reasons, splenectomy could result from certain benign hematological disorders, which are possibly related to subsequent development of hematological malignancies.
34,35 This may partially explain the different risks
of cancer in splenectomy patients with nontrauma and trauma reasons. In vitro and in vivo animal studies have yielded controversial results. Murine models have suggested that splenectomy is associated with increased tumor induction,
31,36 but other studies have shown that splenectomy
may improve the efficiency of cancer treatment in certain tumor host systems.37,38 Prehn performed a review and
concluded that splenectomy on tumor growth is exceedingly complex, and splenectomy enhances or inhibits tumor growth based primarily on the spleen to tumor ratio.8 Table 4 shows the follow-up specific rates for cancer incidence and HR, and splenectomy patients were found to have a consistently significantly higher risk for overall
cancer development among various follow-up times.
Fig. 1 also illustrates that the difference in cumulative incidence curves of all cancers between the splenectomy and
control groups is widening over time. This finding supports that the existence of coincidence of cancer and splenectomy is unlikely. The cancer risk patterns by follow-up years
among diverse cancers varied, and interpreting the discrepancies is difficult.
The strength of this study is its nationwide populationbased design, which increases the generalizability of the
results. The diagnoses of splenectomy and cancer were based on hospital records, which are highly reliable. Our study suggests that splenectomy may need to be employed more conservatively in medical practice in the future, with spleen-preserving techniques used more frequently in trauma care and with splenectomy performed less
frequently in nontrauma indications. However, interpretation of our results should be tempered because of a couple of study limitations. First, information regarding the lifestyle or behavior of patients is lacking in the NHIRD, making it impossible to adjust for health behavior-related factors such as cigarette smoking and alcohol consumption.
These unhealthy habits can increase the risk of cancer39; however, the relationship between health behaviors and splenectomy (and its reasons) remains undetermined. Second, surveillance bias may exist because of increased clinical check-ups in these patients after surgery, which could detect more cancers.
In summary, this population-based retrospective cohort study found significantly increased risks of overall cancer
and certain site-specific cancers in Taiwanese with splenectomy, compared with the general population. The
phenomenon is more prominent when patients had splenectomy because of nontraumatic reasons. The possible
association does not imply causation. The immune factor may play an essential role, and pre-existing hematological disorders may also be related to hematological malignancies.
Further comprehensive investigations should
explore other undetermined mechanisms.
