Cancer risk and cause-specific mortality among people with thymoma in the United States

Introduction

The thymus is essential for generating naïve T-cells and establishing a diverse T-cell repertoire that underpins immune competence and enables recognition of pathogens and tumor antigens¹. As the thymus undergoes age-related involution, thymic output declines and T-cell diversity diminishes, possibly resulting in less effective immunity for new infections and cancer surveillance^{2, 3}. Nonetheless, there is growing evidence that the thymus continues to play an important role in immunity throughout life. For example, a recent study demonstrated that radiographic evidence of the presence of a functioning thymus in adulthood is associated with lower all-cause mortality, reduced lung cancer incidence, and decreased cardiovascular mortality over 12 years of follow-up in the National Lung Screening Trial⁴. Furthermore, Kooshesh et al. found an elevated risk of cancer and increased overall mortality among individuals who underwent thymectomy (surgical removal of the thymus) compared to a control group who underwent other types of cardiothoracic surgery⁵.

Cancers of the thymus are rare, with thymoma representing the most common histological subtype^{6, 7}. Thymomas arise from thymic epithelial cells, which are responsible for directing T-cell maturation^{6, 8}. The function of thymic epithelial cells is disrupted in the setting of thymoma, leading to higher risk of autoimmune conditions including myasthenia gravis, blood disorders, and connective tissue diseases^{9, 10}. Thymoma is also associated with an increased risk of extrathymic malignancies, which might be related to altered immunity, treatment for the thymoma, or environmental and genetic risk factors⁶. For example, data from population-based cancer registries have demonstrated an increased risk for cancer overall (standardized incidence ratios [SIRs] ranging from 1.5 to 3), as well as for non-Hodgkin lymphoma (SIR range 4–7) and prostate cancer (SIR range 1.3–3.0)^11–13. Some single-institution studies have reported the frequency of subsequent cancers without calculating SIRs and have identified lung cancer and other solid tumors as the most common malignancies^{14, 15}. The magnitude and pattern of cancer risk have varied somewhat across studies^11–16, which likely reflects variation in population source, follow-up, and statistical variability due to modest sample sizes (e.g., 192–2171 individuals with thymoma in the population-based studies).

Despite the elevated risk for extrathymic cancers after thymoma, thymoma patients have relatively good prognosis¹⁷. Early-stage thymoma treated with complete thymectomy is associated with 5-year overall survival approaching 90–100% and 10-year survival exceeding 80% for stage I-II disease^{17, 18}. A study of 668 individuals with thymoma using data from the Swedish cancer registry reported that cancer was the leading cause of death (48% of deaths), followed by ischemic heart disease (16%) and pneumonia or chronic obstructive pulmonary disease (COPD, 8%)¹². However, the Swedish study was published a decade ago and only included follow-up through 2004. To our knowledge, there is no population-based study investigating cause-specific mortality among thymoma patients in the United States (US).

National Comprehensive Cancer Network (NCCN) guidelines emphasize a multidisciplinary approach for treatment of thymoma, relying on thymectomy as the primary treatment for early-stage disease and multimodal strategies for more advanced cases¹⁹. As noted above, Kooshesh et al. reported adverse health outcomes among patients who underwent thymectomy for a variety of indications, including thymoma⁵. This study has generated controversy, as some researchers have criticized its findings as incorrectly providing evidence against the value of thymectomy as a treatment for thymoma^{20, 21}. Notably, however, the Kooshesh study compared patients who underwent thymectomy (only a subset of whom had thymoma) with an external control group, rather than comparing between thymoma patients who did or did not undergo thymectomy.

We used data from US cancer registries in the Surveillance, Epidemiology, and End Results program (SEER, 2000–2020) to evaluate the incidence of second malignancies and cause-specific mortality in people diagnosed with thymoma. Further, we investigated the association of thymectomy as treatment for thymoma with subsequent mortality outcomes to inform discussion of the balance of potential benefits and risks of this approach.

Methods

We used population-based cancer incidence and mortality data collected by the National Cancer Institute’s SEER program (https://seer.cancer.gov/). Data were obtained from 17 SEER registries covering Connecticut, Hawaii, Iowa, New Mexico, Utah, Atlanta (Georgia), Detroit (Michigan), Seattle-Puget Sound (Washington), San Francisco-Oakland (California), San Jose-Monterey (California), Los Angeles (California), rural Georgia, greater California, Kentucky, Louisiana, New Jersey, and the Alaska Native population. We limited the SEER data to 2000–2020 due to a marked increase in reported thymoma cases beginning in 2021. This surge reflects changes in how data were reported by cancer registries, which began classifying certain benign or borderline thymomas (previously not reported) as malignant and reportable²². Because these cases were not included before 2021, and due to the lack sufficient follow-up time to assess subsequent cancer incidence and mortality in this group, we restricted our analysis to thymomas diagnosed during 2000–2020.

Patients diagnosed with thymoma were identified from data in the cancer registries, based on the International Classification of Diseases for Oncology (third edition) primary site code C37.9 indicating a tumor in the thymus and histology codes 8580–8585. If a person had more than one thymoma diagnosis, only the first was included in the analysis. Individuals with unknown race/ethnicity were excluded because general population expected rates are unavailable for this group.

Surgical treatment for thymoma was classified using SEER codes. Specifically, total thymectomy was defined as complete or radical removal of the thymus (surgery of primary site codes 40 and 60), and partial thymectomy was defined as partial removal of the thymus or local excision (codes 20–30 and 50). Patients with codes 00 were classified as having no thymectomy performed, and those with codes 90 or 99 as unknown surgery status.

For analysis of subsequent primary cancers among individuals with thymoma, follow-up began two months after thymoma diagnosis to minimize detection bias related to synchronous malignancies. For cause-specific mortality analyses, follow-up commenced at the time of thymoma diagnosis. Individuals were followed until death, loss to follow-up, or the end of the study period, whichever occurred first. Primary endpoints included the incidence of subsequent primary cancers (based on SEER cancer diagnoses) and cause-specific mortality (based on information on the underlying cause of death obtained from death certificates). Cancer types and causes of death were categorized according to SEER groupings, and categories that had zero cases were combined into a miscellaneous category in each analysis.

Standardized incidence ratios (SIRs) for subsequent primary cancers and standardized mortality ratios (SMRs) for cause-specific mortality were calculated using SEER*Stat software (version 9.0.42.0; National Cancer Institute). Expected numbers of cancer cases and deaths were based on general population rates stratified by sex, age (5-year intervals), race/ethnicity, and calendar year (5-year intervals). SIRs and SMRs were calculated as the ratio of observed to expected events, with 95% confidence intervals (CIs) derived assuming a Poisson distribution.

Subsequent thymoma was not included as an outcome in the SIR analysis as it is not possible to rule out that such cases were recurrences of the first diagnosed thymoma. In addition to presenting the SIR for overall cancer (excluding cancers of the thymus), we present an SIR for the combined group of subsequent solid cancers (excluding cancers of the thymus). We present SMRs related to total cancer, all cancers except thymus (based on excluding deaths from “other endocrine cancers including thymus”), which is the category corresponding to death from thymoma), all causes other than cancer, and infectious diseases (including septicemia, pneumonia and influenza, and other infectious and parasite diseases).

Several additional analyses were conducted. First, we conducted sensitivity analyses excluding thymoma patients who did not have diagnostic confirmation. Second, we estimated risk for total subsequent cancer and all-cause mortality according to latency since thymoma diagnosis (<12 months, 12–59 months, 60–119 months, and ≥120 months). Similar analyses were conducted for subsequent thyroid cancer to assess whether the elevated risk could be explained by latency effects of radiotherapy, and for cause-specific mortality from infectious diseases to evaluate whether the increased risk might be attributable to immune disruption associated with thymoma. Third, for subsequent mortality outcomes with at least 15 observed events, we evaluated SIRs or SMRs according to thymectomy status (total, partial, and none). We then used multivariable Poisson regression to estimate mortality rate ratios, comparing individuals who had total or partial thymectomy with those who have no thymectomy as the reference, adjusting for age at diagnosis, sex, and thymoma stage and histological subtype.

All statistical tests were two-sided. Bonferroni-corrected significance thresholds were applied for analyses of SIRs for subsequent cancers (p<0.0015) and SMRs for cause-specific mortality (p<0.0029). We also note SIRs and SMRs that met nominal significance (p<0.05), and all other analyses used this significance level.

Results

A total of 3051 individuals with thymoma diagnosed during 2000–2020 were included in our analysis (Table 1). Most individuals (80.3%) were aged 40–79 years at diagnosis, while pediatric and adolescent cases (age 0–19 years) were rare (0.6%). More than half of the cohort was non-Hispanic White (54.9%), with the next most common groups being non-Hispanic Asian or Pacific Islander (18.4%) and non-Hispanic Black (15.6%). Among all individuals with thymoma, 29.8% were diagnosed with localized thymoma, 34.8% with regional disease, and 15.0% with distant disease; the remaining 20.3% were indicated as having unknown stage. Histologically, thymoma of an unspecified type was the most common (31.1%), followed by types B1-B2 (27.6%), AB (16.9%), B3 (15.9%), and A (8.5%).

As shown in Table 2, among this cohort, 393 subsequent primary cancers were observed, yielding a significant elevation in overall cancer incidence (SIR=1.61; 95% CI=1.46–1.78; p<0.001). Similar excess risk was observed for all solid tumors combined (SIR=1.63, 95%CI=1.46–1.81; p<0.001). Site-specific analyses demonstrated significantly elevated incidence for several malignancies, including lung cancer (SIR=2.06, 95% CI=1.60–2.61), soft tissue sarcomas including cardiac tumors (SIR=5.46; 95% CI=2.36–11.0), urinary bladder cancer (SIR=2.13, 95%CI=1.38–3.14), and thyroid cancer (SIR=4.03, 95%CI=2.35–6.46). Other types of cancer that met the nominal significance level but not the Bonferroni significance level including stomach cancer (SIR=2.55, 95%CI=1.27–4.57), prostate cancer (SIR=1.35, 95%CI=1.01–1.78), and melanoma (SIR=2.08, 95%CI=1.30–3.15). SIRs were essentially the same when we excluded patients without histologic confirmation of their thymoma diagnosis (Table S1).

We evaluated SIRs among individuals with thymoma as a function of time since thymoma diagnosis (Table S2). Overall cancer incidence appeared most strongly elevated in the 2–11 months after thymoma diagnosis (SIR=2.09, 95%CI=1.57–2.73) and appeared lower ≥120 months post-diagnosis (SIR=1.59, 95%CI=1.25–2.00), although this trend was not significant (p-trend=0.397). For thyroid cancer, the SIR was strongly elevated in the period 12–59 months after thymoma diagnosis (SIR=4.61, 95%CI=1.99–9.08), and the SIRs did not increase or decrease as a function of time since thymoma diagnosis (p-trend=0.559).

During follow-up, 1,102 deaths were observed among individuals with thymoma, corresponding to significantly elevated mortality compared with the general population (SMR=2.86; 95% CI=2.70–3.04; Table 3). Mortality was significantly elevated for deaths due to cancer (SMR=6.17, 95% CI=5.67–6.70), including deaths from cancers other than thymus malignancies (SMR=1.76, 95%CI=1.49–2.05), and from all causes other than cancer (SMR=1.84, 95% CI=1.69–2.00). Among non-cancer causes, mortality was elevated specifically related to infectious causes (SMR=2.75, 95%CI=2.04–3.64), including septicemia (SMR=2.66, 95% CI=1.49–4.39), pneumonia and influenza (SMR=2.66, 95%CI=1.67–4.03), and other infectious and parasitic diseases (SMR=3.09, 95%CI=1.59–5.39), as well as from deaths due to diseases of the heart (SMR=1.62, 95%CI=1.37–1.90) and COPD (SMR=1.92, 95%CI=1.39–2.58). At the nominal significance level, mortality from nephritis, nephrotic syndrome, and nephrosis was also elevated (SMR=1.97, 95% CI=1.10–3.24), whereas mortality from Alzheimer’s disease was decreased (SMR=0.36, 95% CI=0.12–0.84). SMRs were largely unchanged when we excluded patients without histologic confirmation of their thymoma (Table S3).

We next examined trends in mortality as a function of time since thymoma diagnosis (Table S4). Overall mortality was most elevated 0–11 months after diagnosis (SMR=4.71, 95%CI=4.10–5.39) and lowest ≥120 months after diagnosis (SMR=2.20, 95%CI=1.90–2.54), with a significant temporal trend (p-trend<0.001). Likewise, mortality from infectious diseases was most strongly elevated 0–11 months after diagnosis (SMR=5.59, 95%CI=2.56–10.6) and lowest ≥120 months after diagnosis (SMR=1.33, 95%CI=0.36–3.41) (p-trend<0.001).

Cause-specific mortality was also evaluated according to thymectomy status (Figure 1, Table S5). SMRs for overall mortality, cancer-specific mortality, and heart disease-related mortality were significantly elevated regardless of thymectomy status. Patients who did not undergo thymectomy had the highest SMRs for overall mortality (SMR=5.74, 95%CI=5.19–6.34) and for death related to cancer (SMR=17.3, 95%CI=15.2–19.6), COPD (SMR=3.20, 95%CI=1.65–5.59), and infectious diseases (SMR=5.84, 95%CI=3.27–9.63), whereas heart disease-related mortality was comparable across groups. Corresponding mortality rate ratios, presented in Figure 2 and Table S5, demonstrate significantly lower overall mortality and mortality due to cancer and infectious diseases in both the partial and total thymectomy groups compared with the no thymectomy group. The inverse associations of partial and total thymectomy with overall mortality and cancer-specific mortality were significant among thymoma patients regardless of stage at diagnosis (Table S6). However, the inverse association between thymectomy and mortality due to infectious diseases was only observed in patients with localized or regional stage thymoma.

Discussion

Based on data for more than 3,000 people with thymoma in the US, our study found that this population has a 1.61-fold higher incidence of cancer and 2.86-fold higher mortality compared with the general population. The cancer types that showed significant risk elevations included lung cancer, soft tissue sarcomas, urinary bladder cancer, and thyroid cancer. Other types of cancer that showed elevations at a nominal significance level included stomach cancer, prostate cancer, and melanoma. Among all causes of death, we observed significantly elevated mortality due to cancer, infectious diseases, diseases of the heart, and COPD.

The thymus plays an essential role in immunity as the site of origin for naïve T-cells, fostering a diverse T-cell receptor repertoire and thereby enabling the immune system to recognize and respond to microorganisms and cancer cells expressing neoantigens²³. Recent studies of large population cohorts and cancer patient shows that better thymic health is associated with lower risk of lung cancer and heart disease, lower all-cause mortality, and enhanced response to cancer immunotherapy^{4, 24}. Another study found that adult patients who received a thymectomy as treatment for a range of medical conditions (including thymoma) had higher overall mortality and cancer risk than a control group of patients who underwent other types of cardiothoracic surgery⁵. Abnormalities in thymic epithelial function that are present in the setting of thymoma may similarly compromise immunity, leading to the elevated risk of additional malignancies observed in prior studies^{6, 11–13, 15, 16}.

In our study, the elevated risks of bladder cancer and lung cancer may be attributable to the impaired immunity following thymoma, as they also have more frequent occurrence in other immunosuppressed populations such as solid organ transplant recipients and (for lung cancer) people with HIV infection²⁵. Other shared risk factors (e.g., smoking) may further contribute to the elevated risks of lung and bladder cancers^{26, 27}. Of interest, we did not observe an elevated risk of lymphoma among individuals with thymoma, although this cancer is associated with immunodeficiency²⁵ and has been associated with thymoma in several prior studies^{11, 12, 16}. This discrepancy might reflect differences in thymoma or lymphoma diagnosis. For instance, two of the prior studies identified thymoma using different diagnostic codes that likely captured a higher proportion of localized and benign cases^{11, 12}. Our cohort, by contrast, may have included more people with advanced or aggressive thymoma, and competing mortality due to thymoma in such patients may have prevented us from observing an excess risk of non-Hodgkin lymphoma.

Cancer-directed radiotherapy is an important risk factor for soft tissue sarcomas²⁸. We considered whether the elevated risk of sarcomas that we observed among individuals with thymoma could be such a complication. However, radiotherapy is unlikely to be the cause of the eight soft tissue sarcomas in our series, as fewer than five cases occurred in the region classified by cancer registries as “upper limb, shoulder, or unspecified trunk,” the part of the body most likely to be exposed to radiation. Additionally, based on SEER treatment data, only a minority of the patients with subsequent soft tissue sarcoma are known to have received radiotherapy. We note that the small numbers of cases according to location and radiation therapy status are suppressed here in accordance with the SEER data use agreement.

In contrast, it is possible that radiotherapy treatment is the explanation for the significantly increased incidence of thyroid cancer after thymoma, as radiation is a known risk factor for malignancies of the thyroid²⁹, which is located immediately superior to the thymus. Sixty percent of people in this study who developed a subsequent thyroid cancer received radiotherapy. Nevertheless, the analysis by latency period did not reveal an increase in SIRs with longer time since thymoma diagnosis, and the strongest elevation of risk appeared to be in the 12–59 months after thymoma diagnosis, which is a shorter interval than typically seen for radiation-induced cancers³¹.

Another potential explanation for the association between thymoma and thyroid cancer is immune dysregulation, because thymoma causes a range of autoimmune disorders, including autoimmune thyroid diseases such as Hashimoto thyroiditis and anti-PIT antibody syndrome^32–34. Moreover, Hashimoto thyroiditis is associated with an increased risk of papillary thyroid cancer (the most common subtype of thyroid cancer)³⁵. Finally, given the strong elevation in thyroid cancer that we observed immediately after thymoma diagnosis, we cannot rule out that the increased incidence of thyroid cancer after thymoma could be due to overdiagnosis, which arises when medical care leads to identification (and often treatment) of cancers that would not have otherwise affected a patient’s life expectancy. Over-diagnosis of thyroid cancer could arise, for example, as a result of surveillance after initial thymoma treatment using chest imaging³⁶. It is similarly possible that other findings of excess cancer risk in this study may arise from over-diagnosis due to imaging (e.g., lung cancer)³⁷.

The excess mortality we observed is likely attributable in part to immune dysregulation associated with thymoma. We identified a 2- to 3-fold higher mortality from infectious diseases, with the greatest elevation in infection-related mortality occurring within the first year after thymoma diagnosis. Good syndrome, in which thymoma is accompanied by hypogammaglobulinemia, is associated with heightened susceptibility to life-threatening bacterial, viral, and opportunistic infections^{38, 39}. In addition, thymoma patients can have normal baseline lymphocyte counts and immunoglobulin levels yet still be vulnerable to severe opportunistic infections because of the presence of autoantibodies against interleukin-23 that block immune activation⁴⁰.

Thymectomy is considered the treatment of choice for limited stage thymoma^{19, 41}. A recent study using the SEER database and National Cancer Database did not find increased mortality among patients with localized thymoma treated with thymectomy compared to the general population or other patients with resectable thyroid or breast cancer²¹. In the present study, we observed elevated mortality among thymoma patients who underwent total or partial thymectomy compared with the general population, consistent with findings from the Kooshesh study⁵. Because thymoma is a major indication for thymectomy, it is possible that the elevated mortality associated with thymectomy in the Kooshesh study could be partially attributable to thymoma, although sensitivity analyses showed consistent patterns after exclusion of patients with thymoma⁵. When we restricted our analysis to patients with localized thymoma, overall mortality was no longer significantly elevated among those who underwent total thymectomy (SMR=1.15, 95%CI=0.86–1.50) and modestly elevated for those with partial thymectomy (SMR=1.48, 95%CI=1.13–1.91), aligning with the results reported by Resio et al²¹. Importantly, neither the Kooshesh nor the Resio study directly addressed whether thymectomy has a positive or negative impact among thymoma patients. Along these lines, a key finding of our study is that, in an analysis adjusting for age, sex, thymoma stage, and histologic subtype. thymoma patients who received partial or total thymectomy had significantly lower all-cause mortality, cancer-specific mortality, and mortality due to infectious diseases compared with thymoma patients who did not undergo thymectomy. These results therefore do not indicate a harm—and, indeed, suggest a benefit—arising from thymectomy among patients with thymoma. Our study thus lends support to NCCN guidelines on the primary role of thymectomy for treatment of thymoma⁴¹.

To our knowledge, this is the largest study of subsequent cancer among people with thymoma and the first study to comprehensively evaluate cause-specific mortality. We leveraged longitudinal, population-based data and applied multivariable Poisson regression to examine the association between surgery and cause-specific mortality—an area of clear clinical relevance given the limited prior literature assessing the impact of thymectomy. In addition, we provide updated estimates of cancer risk following thymoma and conducted additional analyses to strengthen causal inference regarding the etiology of these subsequent malignancies.

Nonetheless, this study has some limitations. First, thymoma is a rare malignancy, so the precision for most analyses was somewhat limited. In addition, we applied a Bonferroni threshold to identify significant increases in cancer and mortality, which is conservative and may have obscured potentially meaningful signals. Second, thymoma has one of the lowest mitotitc indices of any adult tumor and can be clinically diagnosed many years after its development and the appearance of immune abnormalities⁴². Because we measured only the incidence of cancers arising after thymoma diagnosis, we may have missed some cancers related to early undiagnosed thymomas. Third, because SEER does not capture diagnoses of medical conditions other than cancer and lacks data on key risk factors like smoking, we could not construct cumulative incidence curves to guide clinical follow-up or adjust for confounding factors to explain the elevated risks that we observed. Finally, associations between thymectomy and cause-specific mortality should be interpreted with caution, as the apparent protective associations might be confounded by unmeasured tumor characteristics or clinical factors that influence which patients are offered thymectomy, which we were unable to account for.

In conclusion, using population-based data from 17 SEER registries spanning 2000 to 2020, we comprehensively characterized the spectrum of subsequent malignancies and cause-specific mortality among patients with thymoma. We observed significantly elevated risks of lung, thyroid, and bladder cancers as well as soft tissue sarcomas. People with thymoma also experienced increased all-cause mortality and excess mortality from malignancies, infectious diseases, COPD, and heart diseases. These findings underscore the need for cancer prevention and continued efforts to improve survival, especially among people with regional or distant thymomas. Finally, we found that thymoma patients who underwent total or partial thymectomy had lower mortality from all causes of death and deaths from cancers compared with those who did not, supporting current guidelines that this treatment modality should be offered to patients with resectable thymoma.

Supplementary Material