Long‐Term Cancer and Mortality After Thymectomy

ABSTRACT

Background

Long‐term health effects of thymectomy in adults remain uncertain. Previous reports suggested associations with increased mortality, cancer incidence, and autoimmune disease, although these findings may reflect differences in patient selection, surgical indication, or follow‐up duration. We aimed to examine long‐term outcomes following adult thymectomy.

Methods

We performed a propensity score‐matched retrospective study of patients who underwent thymectomy or comparator cardiothoracic procedures, with follow‐up extending to 10 years. Primary outcomes were all‐cause mortality, new‐onset malignancy, and autoimmune disease. Findings were compared with those reported in a recent study that has raised caution regarding thymic excision.

Results

We identified 425 patients who underwent thymectomy and 5401 controls who underwent coronary artery bypass or valve surgery. After matching, 398 patients remained in each group. Thymectomy was not associated with increased mortality at 5 years (HR = 0.63; 95% CI, 0.38–1.04) or 10 years (HR = 0.78; 95% CI, 0.54–1.13). Excluding reclassified thymic cancers, 5‐year extrathymic malignancy incidence was 2.3% versus 1.8% (p = 0.54); no significant differences were observed in immune‐related disease. Mortality was higher among patients with thymic malignancy. Among MG patients, only one event occurred, precluding reliable estimation.

Conclusions

Thymectomy was not associated with excess mortality or malignancy at 10 years compared with matched cardiothoracic controls. Indication‐stratified analyses within the thymectomy cohort showed higher mortality for thymic malignancy, with insufficient events to assess outcomes in MG patients. These findings do not support a generalized survival penalty from thymectomy, though caution is warranted for incidental thymic removal.

In a propensity score‐matched study comparing 398 thymectomy patients with 398 cardiothoracic controls over 10 years, thymectomy was not associated with excess all‐cause mortality (HR 0.78; 95% CI 0.54–1.13) or new malignancy (4.3% in both groups). Indication‐stratified analysis revealed heterogeneity: thymic malignancy carried significantly higher mortality (HR 2.4; p = 0.001), while myasthenia gravis patients showed a favorable trend. These findings challenge prior reports of increased long‐term health risks following adult thymectomy.

1. Introduction

The thymus is a central lymphoid organ responsible for T‐cell maturation and the establishment of autoimmune self‐tolerance [1, 2]. During early life, thymic output is essential for shaping the adaptive immune repertoire [3]. However, with age, the thymus undergoes progressive involution, and its immunologic contribution in adulthood remains uncertain [4, 5]. Because of this decline in its apparent function, the thymus has long been regarded as being dispensable in adults, and thymectomy, whether performed for myasthenia gravis (MG), thymic tumors, or incidentally during cardiothoracic surgery, has traditionally been considered safe with minimal systemic consequences. However, growing evidence challenges this view: thymic involution has been quantitatively linked to age‐related rises in cancer and infectious disease incidence [6], diminished thymic output predicts all‐cause mortality in healthy elderly adults [7], and a 2026 deep‐learning study of routine chest CT in two large prospective cohorts found that preserved thymic health independently predicted lower mortality, reduced lung cancer incidence, and better immunotherapy outcomes [8, 9]. Independently, thymic output has been associated with survival in elderly community‐dwelling adults [10], and genetically determined thymopoiesis predicts T cell response magnitude in infectious disease [11], and affects clinical outcomes after hematopoietic stem cell transplantation [12, 13].

Over the past century, thymic pathology has been most closely associated with MG, in which thymic hyperplasia or thymoma is frequently observed. Experimental and histopathological evidence links these abnormalities to disordered immune regulation and loss of tolerance [14, 15, 16]. Although the clinical benefit of thymectomy in MG is now well established in one randomized and many prospective studies [1, 17], the broader physiological implications of thymic removal for immune homeostasis, malignancy risk, and long‐term survival have remained largely unexplored.

This uncertainty was brought to attention by a 2023 retrospective cohort study by Kooshesh and colleagues [18], who reported higher five‐year mortality and cancer incidence among adults undergoing thymectomy compared with matched controls who had other cardiothoracic operations, with no difference in autoimmune disease incidence. These findings challenged long‐standing assumptions about the innocuous nature of thymectomy in adults and raised concern that loss of thymic function might have broader immunologic and oncologic consequences [1].

To further characterize the long‐term health consequences of thymectomy, we conducted a retrospective cohort study comparing long‐term outcomes against matched cardiothoracic controls. Although the analysis was retrospective, outcomes were predefined and systematic event verification was performed. By extending follow‐up to 10 years, we assessed all‐cause and cause‐specific mortality, incident malignancy, and new‐onset autoimmune disease. We hypothesized that thymectomy does not adversely affect survival or immune‐related outcomes compared with other cardiothoracic procedures.

2. Materials and Methods

2.1. Study Design and Ethical Approval

We conducted a propensity score‐matched retrospective cohort study with prespecified outcomes, eligibility criteria, and a statistical analysis plan established prior to data extraction. Cases were ascertained and validated through manual chart review; mortality outcomes were verified through national and institutional registries, while cancer outcomes were confirmed using the comprehensive hospital registry. Data analysts were blinded to exposure status during case validation and matching. Follow‐up was systematic and obtained from institutional electronic medical records.

The study was intended as a real‐world confirmatory analysis of long‐term outcomes following thymectomy and was approved by the Rambam Health Care Campus Institutional Review Board (IRB No. 0567 22 RMB D). The current cohort partially overlaps with a previously published single‐center analysis from our institution [19] that examined a subset of these patients with shorter follow‐up; the present study extends the observation period to 10 years and incorporates a propensity‐matched cardiothoracic control cohort not included in the prior report.

2.2. Study Population

We included adult patients (≥ 18 years) who underwent thymectomy at Rambam between 1986 and March 2025. Most procedures (87.4%) occurred after January 2000. The control cohort comprised adult patients who underwent coronary artery bypass grafting (CABG) and/or mitral or aortic valve surgery without thymectomy between January 2004 and March 2025. For patients with multiple cardiothoracic procedures, the first eligible surgery defined the index date. This active surgical comparator was selected to minimize healthy‐patient and healthcare‐seeking bias, ensure comparable perioperative exposure (including sternotomy), and enable direct comparison with the design employed by Kooshesh et al. [18].

Patients were excluded from both cohorts if they died within 90 days of the index procedure, to reduce bias from perioperative mortality. Additional exclusions for the control group included a history of myocardial infarction prior to the index procedure and any prior thymectomy.

A flow diagram summarizing cohort derivation is provided (Figure 1). This study was reported in accordance with the STROBE statement for observational research [20].

FIGURE 1.

Cohort derivation and follow‐up outcomes. Flowchart summarizing cohort selection, exclusions, and propensity score matching. After applying eligibility criteria, 425 thymectomy and 5401 control patients were identified, with 398 matched pairs included in the primary analysis. Outcome proportions are shown in the lower panel for descriptive purposes.

2.3. Clinical Variables and Data Sources

Demographic and clinical characteristics, including body mass index (BMI), lipid profile, and comorbidities, were extracted from institutional electronic medical records. BMI and lipid parameters (LDL, HDL, triglycerides) were based on the most recent measurements obtained within 1 year before surgery and up to 30 days postoperatively; when unavailable, the nearest recorded values to the index procedure were used. BMI was calculated as weight (kg) divided by height squared (m²). Hypertension (HTN) and diabetes mellitus (DM) were defined by prior documented diagnoses.

The Charlson Comorbidity Index was calculated from ICD‐9 and ICD‐10 codes using the closest available data to the procedure. Implausible values (BMI < 15 or > 60 kg/m²; LDL < 10 or > 300 mg/dL; HDL < 10 or > 100 mg/dL; triglycerides < 10 or > 1000 mg/dL) were set to missing and replaced with the nearest valid measurement.

2.4. Exposure Classification and Subgroups

The exposure of interest was thymectomy versus control surgery. Among thymectomy patients, the preoperative indication for surgery was categorized as: thymic malignancy, myasthenia gravis (MG), or other indications (no MG, no thymic malignancy). These subgroups were used for within‐cohort comparative analyses. In cases where both thymic malignancy and MG were recorded, the indication was classified as malignancy. Indications were determined based on diagnoses occurring from the period prior to surgery up to 90 days post‐procedure, capturing the clinical context that led to or was directly related to the operation. Histological subtype and Masaoka‐Koga stage were obtained by manual review of pathology reports for patients in the thymic malignancy indication group.

2.5. Outcomes

The primary outcome was all‐cause mortality from 91 days postoperatively through 10 years of follow‐up, with a parallel analysis limited to 5 years.

Secondary outcomes included: First‐ever pathologically or serologically confirmed malignancy diagnosed more than 90 days after the index procedure. Accordingly, recurrence, progression, or subsequent malignancies occurring in patients with a pre‐existing cancer diagnosis were not counted as primary malignancy events. First diagnosis of a new autoimmune condition post‐surgery (identified using ICD‐9 and ICD‐10 codes from institutional records and confirmed by investigator review), and cause‐specific mortality (e.g., cancer‐related, cardiovascular), when available. To ensure accuracy of event classification, all cases of mortality and autoimmune disease were individually reviewed by study investigators (S.S. and M.A.) to ascertain each diagnosis. All pathology reports and discharge summaries were reviewed. Each case was verified for diagnostic validity, timing in relation to surgery, and correct categorization of cause of death or autoimmune phenotype.

2.6. Follow‐Up and Censoring

Patients were followed from the index date until the earliest of outcome occurrence, 10 years post‐surgery, death, or last recorded follow‐up. Patients with less than 5 years of follow‐up were right censored at last known contact. Five‐ and ten‐year survival estimates were obtained from Kaplan–Meier survival curves. Crude rates were also calculated as simple proportions among those with any follow‐up time within each cohort.

2.7. Statistical Analysis

Baseline characteristics were compared between groups using the tableone package in R, with a significance threshold of p < 0.05. Standardized mean differences and p‐values were reported for key demographic and clinical variables. To address baseline differences between groups, 1:1 nearest‐neighbor propensity score matching was performed without replacement. The propensity score was estimated using logistic regression incorporating age and sex. This parsimonious specification was chosen for two reasons: first, to enable direct methodological comparison with Kooshesh et al., who used the same matching variables; and second, to maximize matching efficiency given the sample size and the marked imbalance between groups in cardiovascular risk factors, which limited covariate overlap between the thymectomy and control populations. A caliper of 0.10 on the logit of the propensity score was applied, and exact matching was enforced on sex and age group. To assess the potential impact of residual cardiovascular risk factor imbalance, a sensitivity propensity score matching analysis was performed, incorporating age, sex, diabetes, and hypertension.

Time‐to‐event analyses were performed using Kaplan–Meier curves with log‐rank tests to compare survival between groups. Cox proportional hazards models were used to estimate hazard ratios (HRs) with 95% confidence intervals (CIs), comparing thymectomy patients to controls; these primary between‐group models were unadjusted beyond the propensity‐matched design. For malignancy outcomes, death was additionally modeled as a competing event using cumulative incidence functions and Fine–Gray regression.

For within‐cohort subgroup analyses (thymectomy indications), Cox models were used to compare mortality risks, using the “other indications” group as the reference. The proportional hazards assumption was assessed using the Schoenfeld residuals test. As an additional sensitivity analysis, patients in the thymic malignancy indication group were excluded, propensity score matching was repeated, and mortality and malignancy outcomes were re‐evaluated. All analyses were conducted using R (version 4.4.2).

Large language model‐based tools were used solely for language editing and formatting of the manuscript. These tools did not contribute to study design, data analysis, interpretation, or generation of results.

3. Results

3.1. Study Participants

After applying exclusion criteria, we identified 425 patients who underwent thymectomy (median age: 54 years [IQR 39–66]; 50.4% women) and 5401 control patients who underwent CABG and/or valve surgery (median age: 67 years [IQR 59–75]; 32.6% women). Following propensity score matching (PSM), 398 patients were retained in each group. Post‐matching balance was excellent for the matched variables (SMD 0.01 for age, < 0.001 for sex); full post‐matching baseline characteristics including residual differences in Charlson index, lipids, and comorbidities are provided in Appendix S1. The median follow‐up duration was 3552 days (9.7 years; IQR 1725–3650 days, with the upper quartile reaching the study's administrative censoring limit of 10 years); 82.4% of thymectomy patients and 80.4% of controls had at least 5 years of follow‐up. Additional demographic characteristics are summarized in Table 1.

TABLE 1.

Baseline characteristics of patients undergoing thymectomy and surgical controls.

	Control (n = 5401)	Thymectomy (n = 425)	p
Sex (Male, %)	3638 (67.4)	211 (49.6)	< 0.001
Age (median [IQR])	67.00 [59.00, 75.00]	54.00 [39.00, 66.00]	< 0.001
Age groups (%)			< 0.001
0,30	35 (0.6)	60 (14.1)
30,40	99 (1.8)	56 (13.2)
40,50	404 (7.5)	55 (12.9)
50,60	1027 (19.0)	101 (23.8)
60,70	1649 (30.5)	91 (21.4)
70,100	2187 (40.5)	62 (14.6)
Charlson index (median [IQR])	5.00 [4.00, 8.00]	8.00 [3.00, 11.00]	< 0.001
LDL levels (median [IQR])	74.00 [51.60, 102.00]	97.15 [70.00, 126.00]	< 0.001
HDL levels (median [IQR])	34.90 [27.00, 43.60]	43.00 [32.00, 53.00]	< 0.001
Triglyceride levels (median [IQR])	102.00 [75.00, 153.60]	111.50 [75.00, 159.75]	0.277
BMI (median [IQR])	28.50 [25.50, 32.00]	27.20 [24.00, 31.50]	< 0.001
BMI ≥ 30 (%)	1652 (38.0)	78 (31.1)	0.034
Diabetes (%)	2425 (44.9)	72 (17.0)	< 0.001
Hypertension (%)	4174 (77.3)	154 (36.4)	< 0.001
Myasthenia Gravis before index (%)	1 (0.0)	86 (20.2)	< 0.001

Note: Values are presented as n (%) for categorical variables and median [IQR] for continuous variables. Prior to matching, patients undergoing thymectomy were younger, included a higher proportion of women, and had substantially lower rates of cardiovascular risk factors and comorbidities than controls. Despite this, the Charlson Comorbidity Index was higher in the thymectomy group, reflecting the contribution of disease‐specific conditions such as thymic malignancy and myasthenia gravis.

3.2. Mortality Outcomes

In the unmatched sample, crude 10‐year mortality was higher among controls (25.4%) than thymectomy patients (12.6%). The Kaplan–Meier survival curves in the unmatched cohort similarly demonstrated lower survival among controls (Appendix S2). After matching, 10‐year mortality rates were 15.8% in controls and 13.3% in thymectomy patients. The absolute total number of deaths in the thymectomy group remained unchanged (n = 54), reflecting the exclusion of younger patients without matched controls. Kaplan–Meier survival curves after matching showed no significant differences in survival between groups (Figure 2). In Cox proportional hazards models, the adjusted hazard ratio (HR) for all‐cause mortality associated with thymectomy was 0.78 (95% CI: 0.54–1.13; p = 0.19) over 10 years, and 0.63 (95% CI: 0.38–1.04; p = 0.07) at 5 years. The proportional hazards assumption was not violated (Schoenfeld test p = 0.12). Although the point estimates suggested a lower mortality risk in the thymectomy group, the differences were not statistically significant. Cause of death was recorded in all available patients (Appendix S3). In a pre‐specified calendar‐period sensitivity analysis restricted to thymectomy procedures performed after 2004 (n = 307 matched pairs), 10‐year all‐cause mortality was 12.9% in the thymectomy group versus 16.8% in matched controls (HR = 0.76; 95% CI, 0.50–1.14; p = 0.19). The 5‐year hazard ratio was 0.51 (95% CI, 0.28–0.94; p = 0.029), indicating that the main finding of no excess mortality is robust to the exclusion of the pre‐2004 thymectomy cohort; if anything, restriction to the contemporary era strengthened the direction of effect. Additional sensitivity analysis using an age‐ and sex‐adjusted Cox proportional hazards model in the full cohort yielded consistent results (Appendix S4).

FIGURE 2.

Kaplan–Meier survival curves comparing overall survival between thymectomy and matched control cohorts over 10 years of follow‐up. The main plot displays cumulative survival probability from the index procedure for both groups. The inset shows the same survival curves using an expanded y‐axis (85–100% survival) to facilitate visualization of small differences between groups. No significant difference in overall survival was observed (HR = 0.78, 95% CI 0.54–1.13; p = 0.19).

Subgroup analysis by indication within the unmatched thymectomy cohort revealed significant heterogeneity in outcomes (p < 0.001; Figure 3). Event counts were: thymic malignancy (n = 140, 28 deaths [20.0%]); MG (n = 56, 1 death [1.8%]); other indications (n = 229, 25 deaths [10.9%]). Thymic malignancy was associated with increased mortality (HR = 2.42, 95% CI: 1.41–4.17; p = 0.001). In contrast, preoperative MG showed a non‐significant trend toward lower mortality (HR = 0.30, 95% CI: 0.03–1.14; p = 0.082, Firth‐corrected); this result was based on a single event and should be interpreted with caution. No meaningful difference in mortality was observed by sex within the thymectomy group; however, in the control group, female sex was associated with lower survival (Appendices S5 and S6). Increasing age was consistently associated with higher mortality across both cohorts (Appendices S4 and S7).

FIGURE 3.

Kaplan–Meier survival analysis within the thymectomy cohort (before matching), stratified by preoperative indication (thymic malignancy, myasthenia gravis (MG), or other non‐malignant indications). Patients who underwent thymectomy for thymic malignancy had significantly lower survival compared with those without malignancy or MG (p = 0.00017). In contrast, patients with preoperative MG showed a trend toward better survival, though not statistically significant. The accompanying forest plot displays hazard ratios for mortality by indication, using the ‘other indications’ group as the reference, confirming that thymic malignancy was associated with a more than twofold higher risk of death (HR = 2.42, 95% CI 1.41–4.17, p = 0.001), whereas MG before thymectomy was not significantly associated with increased mortality (Firth‐corrected HR = 0.30, 95% CI 0.03–1.14, p = 0.082; standard Cox HR = 0.20, 95% CI 0.03–1.50; p = 0.117).

3.3. Incidence of Malignancy

The observed 10‐year incidence of malignancy was identical in the matched thymectomy and control groups (4.3%; n = 17 in each group). Kaplan–Meier analysis of time to first malignancy showed no significant difference between groups (p = 0.67; Figure 4). At 5 years, the incidence of malignancy was 3.5% (14/398) in the thymectomy group and 1.8% (7/398) in controls. Among the 14 malignancies observed in thymectomy patients, 5 cases (35.7%) were thymic malignancies that arose in individuals initially classified with non‐malignant thymoma (Appendix S8). Excluding reclassified cases, extrathymic malignancy incidence was 2.3% (9/398) in the thymectomy group versus 1.8% (7/398) in controls at 5 years (p = 0.54), and 3.0% versus 4.3% at 10 years. These findings indicate that the early numerical excess was driven by reclassified thymic disease rather than excess extrathymic oncogenesis. A competing‐risk analysis treating death as a competing event yielded similar findings (Gray's test p = 0.054; Fine–Gray sHR 2.38, 95% CI 0.96–5.89; p = 0.061), indicating that differential mortality did not materially affect the interpretation of malignancy incidence. In a sensitivity analysis incorporating diabetes and hypertension into the propensity score model, 372 matched pairs were retained. Balance was excellent for age, sex, diabetes, and hypertension (all SMDs < 0.03). Results were consistent with the primary analysis, with no increase in malignancy risk at 5 years (HR 1.67, 95% CI 0.71–3.91; p = 0.237) or 10 years (HR 0.97, 95% CI 0.49–1.90; p = 0.926).

FIGURE 4.

Kaplan–Meier curves showing the cumulative probability of remaining cancer‐free during the 10‐year follow‐up period among thymectomy and matched control patients. The main panel illustrates the overall cancer‐free survival after surgery, while the inset zooms into the 85%–100% range, highlighting the early separation between the curves. The incidence of new malignancies did not differ significantly between groups (HR = 1.16, 95% CI 0.59–2.27; p = 0.67). Notably, nearly all cancer events in the thymectomy group occurred within the first 5 years, with no new cases observed beyond 5.6 years of follow‐up.

In an additional sensitivity analysis excluding patients in the thymic malignancy indication group (n = 140), propensity score matching was repeated and outcomes were reanalyzed. Although a numerical excess of malignancies remained at 5 years (HR 2.64, 95% CI 0.94–7.40; p = 0.065), no statistically significant increase was observed, and the association attenuated by 10 years (HR 1.76, 95% CI 0.78–3.99; p = 0.17). Among the 140 patients in the thymic malignancy indication group, histological subtype and Masaoka‐Koga stage distributions are summarized in Appendix S9.

3.4. Immune‐Related Disease

Within 10 years, immune‐related diseases were diagnosed in 4.5% (18/398) of thymectomy patients and 2.5% (10/398) of controls. Among thymectomy patients, 6 cases (33.3%) were MG. After excluding MG, the incidence of other immune‐related diseases was 3.0% (12/398) in thymectomy patients and 2.5% (10/398) in controls. At 5 years, the incidence of immune‐related diseases was 3.8% (15/398) in the thymectomy group and 1.8% (7/398) in controls. Kaplan–Meier curves showed no significant differences in immune‐related disease incidence between groups (p = 0.084 at 5 years, p = 0.13 at 10 years) (Appendices S10 and S11).

4. Discussion

In this matched, retrospective cohort study, we evaluated the long‐term health consequences of thymectomy versus comparator cardiothoracic procedures. Across 398 matched pairs with up to 10 years of follow‐up, thymectomy was not associated with higher all‐cause mortality, cancer incidence at 10 years, or increased risk of immune‐related disease. Five‐year cancer incidence was numerically higher after thymectomy, driven in part by thymic malignancies reclassified during follow‐up, but differences were not statistically significant; immune‐related diagnoses were also similar between groups. Subgroup analyses suggested heterogeneity by indication: thymectomy for thymic malignancy conferred higher mortality, whereas outcomes in the preoperative MG group were difficult to assess due to very few events. Our primary outcome, overall survival, did not differ significantly after thymectomy at 5 or 10 years (adjusted HR at 10 years, 0.78; 95% CI, 0.54–1.13).

These findings contrast with those of Kooshesh et al. [18], who reported higher 5‐year all‐cause mortality after thymectomy than after matched cardiothoracic surgery (8.1% vs. 2.8%; relative risk, 2.9; 95% CI, 1.7–4.8) and greater cancer mortality with longer follow‐up (9.0% vs. 5.2% compared with the U.S. population). To allow direct comparison, we also analyzed outcomes at the 5‐year time point in our cohort. At this interval, the adjusted hazard ratio for all‐cause mortality was 0.63 (95% CI, 0.38–1.04), indicating no excess mortality after thymectomy, and the crude 5‐year incidence of malignancy was 3.5% vs. 1.8% in controls (p = 0.053). Extending follow‐up to 10 years yielded overall lower mortality and malignancy rates after thymectomy. In contrast, in the 5‐year Kooshesh cohort, both outcomes were higher after thymectomy [18].

These comparisons underscore how follow‐up duration shapes the interpretation of thymectomy outcomes: our 5‐year and 10‐year analyses yield consistent conclusions and align with the extended follow‐up reported in prior studies [21]. Our findings are consistent with a recent large‐scale analysis using the Surveillance Epidemiology and End Results (SEER) and National Cancer Database (NCDB) databases, which examined more than 2400 patients who underwent total or partial thymectomy for localized or small thymoma. That study found no excess 5‐year mortality or cancer‐related death compared with the general U.S. population or with propensity‐matched cohorts undergoing thyroid lobectomy or partial mastectomy. Survival did not differ between partial and total thymectomy, and death from secondary malignancy was rare and comparable to rates observed after resected breast or thyroid cancer [22]. These results parallel our own observations and further strengthen the conclusion that thymectomy in adults does not confer increased mid‐term mortality or oncologic risk. Notably, the study by Resio et al. [22] addressed a complementary clinical question by comparing thymectomy outcomes with demographic‐adjusted population expectations rather than with a surgical control cohort. While population‐based and surgical‐control comparisons each have distinct methodological strengths and limitations, the concordant findings across these different approaches strengthen the evidence that thymectomy is not associated with a substantial increase in mid‐term mortality or cancer‐related death. Subsequent commentaries have highlighted potential residual confounding in the Kooshesh study [18], including concerns about the matching strategy, subgroup construction, endpoint ascertainment, and the timing of divergence in event curves [23, 24, 25].

Our indication‐stratified analyses revealed heterogeneity by indication: mortality was higher when thymectomy was performed for thymic malignancy, whereas the point estimate for MG patients was lower (HR = 0.30; 95% CI, 0.03–1.14; Firth‐corrected) compared with other thymectomy indications. It is important to note that this HR reflects a within‐thymectomy comparison of MG patients versus patients with other indications who also underwent thymectomy and does not represent a comparison of MG patients who underwent thymectomy against those managed without surgery. The estimate should be interpreted with caution given the wide confidence interval and single event in the MG group. The MGTX trial, which did address the question of thymectomy versus medical management in MG, established clinically meaningful benefits including reduced disease severity, fewer exacerbations, and lower immunosuppressant exposure at 3 years, with durability through 5 years; survival and malignancy outcomes were not assessed in that trial [26].

By 10 years, cumulative cancer incidence did not differ between thymectomy and controls in our matched analysis, although 5‐year events were higher after thymectomy and included reclassified thymic malignancies among cases initially classified as non‐malignant. A sensitivity analysis excluding patients in the thymic malignancy indication group yielded qualitatively similar results, suggesting that the observed findings were not explained solely by the inclusion of patients with thymic epithelial neoplasms. Kooshesh et al. reported a higher 5‐year cancer risk post‐thymectomy (7.4% vs. 3.7%; RR, 2.0; 95% CI, 1.3–3.2) and described more aggressive phenotypes on chart review, although those subsets were unmatched and therefore vulnerable to selection bias. Small contemporary series in MG are mixed, with limited power for extrathymic events; current literature does not resolve whether the 5‐year signal reflects causal biology, early detection bias, or case‐mix [27].

We observed no difference in incident immune‐related disease between cohorts over 10 years, including after exclusion of post‐thymectomy MG diagnoses. Kooshesh et al. [18] found no overall difference but reported higher autoimmune incidence when excluding patients with preoperative infection, cancer, or autoimmune disease (12.3% vs. 7.9%; RR, 1.5; 95% CI, 1.02–2.2). Population‐based data from Taiwan likewise suggested a higher incidence of autoimmune disease after thymectomy, particularly for systemic lupus erythematosus, Hashimoto's thyroiditis, and autoimmune hemolytic anemia, although these analyses could not fully account for indication and surveillance intensity [28]. Case series in thymic epithelial tumors show complex bidirectional associations with autoimmunity that are not clearly modified by prior thymectomy, reinforcing the role of underlying tumor biology [25]. Recent critical reviews synthesize these mixed findings and advise caution in extrapolating from heterogeneous, observational cohorts [29]. Two recent single‐center studies are consistent with our findings. Khateb et al. found no increase in extrathymic cancer or autoimmune disease after thymectomy in non‐thymomatous MG patients [30]. Hanson et al. observed no excess malignancy in an exploratory surgical cohort, but reported a higher incidence of new autoimmune diagnoses [31].

These findings must be interpreted in the context of a growing body of evidence establishing the immunological significance of the adult thymus. Epidemiological, mechanistic, and imaging‐based studies have consistently demonstrated associations between thymic function and cancer incidence, infectious disease outcomes, immune reconstitution, and all‐cause mortality in aging populations [6, 7, 8, 9, 10, 11, 12, 13, 19, 27]. Our study addresses the specific question of whether surgical thymectomy, when clinically indicated, is associated with adverse long‐term outcomes and does not refute the broader biological importance of the thymus. Taken together, our data do not support a generalized survival penalty from thymectomy when compared with matched cardiothoracic controls; risks appear contingent on indication. For patients with acetylcholine receptor antibody‐positive generalized MG, the totality of evidence including the MGTX trial supports the clinical benefit of thymectomy, and our indication‐stratified analyses within the thymectomy cohort showed no statistically significant increase in mortality in the MG subgroup, though this comparison was underpowered [17]. Pediatric cardiac surgery provides a further point of comparison. Incidental thymectomy is performed routinely in children undergoing open‐heart surgery. A systematic review by Cavalcanti et al. reported that thymectomized children show alterations in T cell subsets, including reduced naive CD4+ T cells and recent thymic emigrants, and that these changes may persist into later childhood [32]. A clinically significant immunodeficiency has nonetheless not been described in these children. The thymus is most active in early life, and the absence of a clinical phenotype after its removal at this stage is consistent with our finding that thymectomy in adults was not associated with increased mortality or cancer death.

Overall, thymectomy was not associated with excess mortality at 5 or 10 years compared with matched controls. Outcomes varied by indication, with higher mortality in thymic malignancy and insufficient events to assess MG. Early cancer incidence was modestly higher but did not differ by 10 years. These findings support the absence of a clear increased risk from thymectomy when clinically indicated, while emphasizing that incidental or non‐essential thymic removal should be approached with caution. Preservation of thymic tissue, when feasible, remains prudent pending further long‐term evidence.

This was a single‐center, retrospective analysis subject to residual confounding and misclassification. Matching was restricted to age and sex; important unmeasured variables including smoking history, comorbidity severity, tumor stage, adjuvant therapy, and functional status may differ between groups and are not accounted for. The use of cardiothoracic surgical controls, while designed to minimize healthy‐patient bias and enable comparison with Kooshesh et al. [18], may introduce residual confounding related to differences in underlying disease profiles that are not fully addressed by matching on age and sex. Cancer ascertainment may be influenced by differential imaging intensity: thymectomy patients, particularly those with thymic tumors, typically receive more intensive oncologic surveillance, which may inflate early detection of malignancy in that group. Similarly, MG patients receive more intensive immunological follow‐up than cardiac surgery controls, which may increase detection of autoimmune diagnoses in the thymectomy group independent of any true biological effect of thymic removal. Subgroup analysis for MG was underpowered due to very few events, limiting interpretation. Post hoc power analysis (Schoenfeld formula) indicated that with 116 total mortality events across 10 years, the study had 26% power to detect the observed HR of 0.78 and was powered to detect HRs of ≥ 1.68 or ≤ 0.60 at 80% power; the observed effect lies within the indeterminate zone, and a true modest excess or deficit in mortality cannot be excluded. The thymectomy cohort spans 1986–2025 while controls were enrolled from 2004 onward, introducing potential calendar‐period bias; a sensitivity analysis restricted to post‐2004 procedures yielded consistent conclusions. Finally, surgical technique, cancer diagnostics, and immunosuppression protocols also evolved over this period.

Author Contributions

Shahar Shelly: conceptualization, supervision, writing – original draft, methodology, investigation. Mor Ani: investigation, writing – original draft. Keren Miron: data curation, writing – review and editing. Yuebing Li: writing – review and editing. Christopher Klein: writing – review and editing. Roni Ramon‐Gonen: formal analysis, writing – original draft, visualization, methodology, conceptualization, investigation. Gil I. Wolfe: writing – review and editing. Henry Kaminski: writing – review and editing.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Appendix S1: Baseline characteristics of patients undergoing thymectomy and surgical controls after matching for age and sex.

Appendix S2: Kaplan–Meier survival curves before matching, comparing thymectomy and control cohorts. Before propensity score matching, thymectomy patients (blue line) exhibited substantially higher 10‐year survival compared with surgical controls (orange line) (p < 0.0001). Shaded areas represent 95% confidence intervals, and the table below shows the number of patients at risk over time.

Appendix S3: Death reasons within 10 years 63 patients died within 10Y, 29 died in Rambam. The leading causes of death in the control cohort were multiple organ failure (27.6%) and heart failure (20.7%), followed by sepsis (17.2%) and respiratory failure (13.8%). Less frequent causes included undetermined (10.3%), intracerebral hemorrhage (6.9%), and stroke (3.4%). Women predominated across most causes, accounting for most deaths from multiple organ failure, heart failure, and sepsis. In contrast, male deaths were more evenly distributed across respiratory failure, hemorrhagic stroke, and undetermined causes. Overall, mortality was largely driven by systemic and cardiovascular complications, with a modest sex imbalance favoring higher rates among females in multi‐organ and septic etiologies.

Appendix S4: Study Sensitivity Analysis In the full cohort, we performed an adjusted Cox proportional hazards regression to estimate the independent effect of thymectomy on mortality while accounting for confounding by age and sex. This approach leverages the full sample size and controls for key covariates that differ between groups, ensuring that the observed association reflects an age‐ and sex‐adjusted hazard ratio rather than crude group differences. In the full cohort, adjusted Cox regression showed that thymectomy was associated with a significantly lower hazard of all‐cause mortality compared with surgical controls (HR = 0.74, 95% CI: 0.56–0.98; p = 0.035). Older age was strongly associated with increased mortality (HR = 1.07 per year, 95% CI: 1.06–1.07; p < 0.001). Male sex was associated with a non‐significant trend toward lower mortality compared with females (HR = 0.91, 95% CI: 0.82–1.02; p = 0.092). The estimated hazard ratio for thymectomy in the matched cohort (HR = 0.78) was similar in magnitude, supporting the robustness of the observed protective association across analytical approaches.

Appendix S5: Kaplan–Meier survival curves for the control cohort after matching, stratified by sex. Male patients demonstrated slightly higher 10‐year survival compared with females (log‐rank p = 0.036). Shaded areas represent 95% confidence intervals, and the table below shows the number of individuals at risk at each time point.

Appendix S6: Kaplan–Meier survival curves after matching, stratified by treatment group and sex. Matched control females (blue line) showed lower long‐term survival compared with the other subgroups, while thymectomy patients demonstrated numerically higher survival in both sexes (log‐rank p = 0.077). Shaded areas represent 95% confidence intervals, and the table below displays the number of patients at risk over time.

Appendix S7: Kaplan–Meier survival curves for the control cohort, stratified by age group. Patients aged 54 years and older (blue line) exhibited significantly lower long‐term survival compared with younger patients (under 54 years; orange line) (p < 0.0001). Shaded areas represent 95% confidence intervals, and the table below shows the number of individuals at risk at each time point.

Appendix S8: Malignancies identified within 5 years after surgery in the matched thymectomy and control cohorts.

Appendix S9: Histological subtype and Masaoka‐Koga stage distribution among patients in the thymic malignancy indication group (n = 140) Patients were classified into the thymic malignancy indication group based on preoperative diagnostic coding suggestive of thymic or mediastinal malignancy. “No tumor” indicates cases in which no neoplastic lesion was identified on final pathological review despite a preoperative malignant indication.

Appendix S10: Kaplan–Meier curves showing immune‐related disease‐free survival up to 10 years after surgery, stratified by treatment group. Thymectomy patients (blue line) and matched surgical controls (orange line) showed comparable rates of new‐onset immune‐related diseases during follow‐up (p = 0.13). Shaded areas represent 95% confidence intervals, and the table below indicates the number of patients at risk over time.

Appendix S11: Autoimmune diseases following surgery in matched thymectomy and control groups.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.