Management and Long-Term Outcomes of Advanced Stage Thymoma in the United States

Abstract

Background.

Thymomas are rare tumors, with limited data regarding treatment of advanced stage disease. Although surgical resection is the mainstay of treatment, the role of additional therapy remains controversial. Our objectives were to describe treatment strategies for stage III/IV thymoma in the United States and compare survival outcomes among treatment approaches.

Methods.

We identified Masaoka stage III/IV thymoma reported in the National Cancer Database between 2004 and 2016. Frequencies of treatment with surgery, chemotherapy, radiation, and combinations were calculated. Five-year overall survival was compared using the Kaplan-Meier method and log-rank test. Risk-adjusted proportional hazards modeling compared mortality between treatment regimens.

Results.

A total of 1849 patients were identified (1108 stage III, 741 stage IV). Among stage III patients, 83.8% underwent resection (± other modalities) compared with 60.2% of stage IV. Surgery plus radiation was the most common regimen for stage III (32.6%), and nonsurgical treatment (definitive chemotherapy and/or radiation) was the most common for stage IV (36.4%). Overall 5-year survival was 70.3% for stage III and 58.5% for stage IV. In risk-adjusted analysis, surgery plus radiation had the lowest mortality (hazard ratio 0.41, 95% confidence interval 0.30–0.55). Patient age, tumor size, metastases, and non-academic treating hospital were associated with mortality.

Conclusions.

Current treatment regimens for advanced stage thymoma vary significantly. Regimens that include surgical resection are most common and are associated with superior outcomes. Patients selected to have surgery as primary treatment had the best survival. Adjuvant radiation treatment is associated with better survival and should be considered in patients who undergo resection.

Thymomas are relatively rare tumors, with an estimated incidence of 1.5 to 2.7 per million per year.^1–3 These cancers represent the most common malignancy of the anterior mediastinum, however, and the incidence of the diagnosis has increased in the past decade.¹ Surgical resection is the mainstay of treatment, and completeness of resection is the most important prognostic factor.^4–6 Some thymomas present in more advanced stages, for which additional local or systemic therapy may be warranted. Because of its low incidence, there are limited data regarding treatment of thymoma in its advanced stages.

Prospective data investigating treatment of advanced stage thymomas is especially lacking. Previous studies have attempted to address various questions regarding multimodality treatment of advanced thymoma. A number of large retrospective series have examined the use of adjuvant radiation therapy for advanced stage thymoma without a clear consensus as to the efficacy of this treatment strategy.^7–11 In addition, 2 small prospective series have suggested higher rates of complete resection and improved survival using neoadjuvant radiation or chemoradiation.^12,13 Although these studies provide some insight into multimodality treatment of advanced thymoma, the series are significantly limited by sample size, given the rarity of these tumors. Further, these data represent the experience of a limited number of institutions and may not accurately represent the broader current clinical practice in the United States.

Given the uncertainty regarding optimal treatment of advanced stage thymoma, and particularly the uncertain role of both neoadjuvant and adjuvant therapies, marked differences may exist in the approach to treatment of this rare disease. We sought to utilize a large national cancer registry to (1) describe treatment strategies currently used for Masaoka stage III and IV thymoma in the United States, and (2) compare overall survival outcomes among these different treatment modalities.

Patients and Methods

Data Source and Patients

The National Cancer Database (NCDB) is a joint venture by the American College of Surgeons and the American Cancer Society, and is the largest cancer registry in the world. It is a hospital-based registry, with trained clinical registrars abstracting data from Commission on Cancer accredited facilities and includes >70% of cancers diagnosed in the United States.¹⁴

We identified patients with Masaoka stage III or IV thymoma from the NCDB between 2004 and 2016 using International Classification of Diseases for Oncology, 3rd ed., histologic codes for thymoma (8580–8585). Cases of thymic carcinoma (8586) were excluded. As Masaoka staging criteria is not directly available in the NCDB, a variable indicating tumor extension was used to derive the appropriate stage, similar to previous studies (Appendix 1).⁷ Stage III was defined as “extension into adjacent organs/structures in the mediastinum” and stage IV as “further contiguous extension” or the presence of positive lymph nodes or distant metastases. Cases with missing data for extension and metastasis were excluded. We also excluded patients with multiple cancers and those receiving palliative treatment.

Patient and Treatment Variables

Patient demographic and clinical variables were chosen a priori for inclusion in analysis. Patient factors included age, sex, race/ethnicity, insurance status, and Charlson comorbidity score. In addition to Masaoka stage and presence of metastases, tumor characteristics recorded were histologic subtype and tumor size, when available. Year of diagnosis and academic status of treating hospital were also evaluated.

The NCDB records all available information regarding the “first course of treatment”, which includes all planned and delivered treatment modalities prior to any possible recurrence or progression. Treatments recorded include surgical therapy, radiation therapy, and chemotherapy, and the relative times of each are indicated in the NCDB. We first divided patients into surgically treated and nonsurgically treated. Among nonsurgically treated patients, patients received either no treatment or radiation and/or chemotherapy. Surgically treated patients included groups of patients treated with surgery alone, surgery plus radiation, surgery plus chemotherapy, or surgery plus chemoradiation. Initial overall analysis did not distinguish between timing of therapy (ie, surgery plus radiation regardless of neoadjuvant or adjuvant). However, subsequent subanalysis investigated differences in timing of radiation and use of induction therapy. Completeness of resection (R0 resection vs R1/R2 resection) was also investigated as an outcome in the subanalysis, as well as a stratifying variable for survival analysis.

Statistical Analysis

Frequencies of each treatment modality were determined first overall, and then stratified by Masaoka staging. Five-year overall survival for each treatment modality was determined using the Kaplan-Meier method and compared using the log-rank test. Cox proportional hazards modeling was then used to compare survival among the different treatment modalities with adjustment for all variables outlined above. Additionally, another model similarly compared survival specifically among patients treated surgically, using the same covariates as the previous model and adding “completeness of resection”. Finally, among patients without metastasis who were treated surgically, overall survival was compared between patients who received neoadjuvant and/or adjuvant therapy, or surgery alone.

Patients treated with both surgery and radiation were separated into neoadjuvant and adjuvant groups. Survival between these 2 groups was compared using the Kaplan-Meier method. Subsequently, in a subset of patients with stage III/IV without metastases, we investigated the rate of R0 resection for patients receiving induction therapy followed by surgery compared to primary surgery alone using an unadjusted χ² test and adjusted multivariable logistic regression. In addition, we evaluated the association between adjuvant radiation therapy and survival stratified by completeness of resection.

Results

We identified a total of 1849 patients with Masoka stage III/IV thymoma reported between 2004 and 2016 (1108 stage III and 741 stage IV) (Figure 1). Average age was 57.4 years, with an equal distribution between the sexes (Table 1). These patients tended to have more aggressive histologic subtypes (21.3% type B3) and large tumors (29.0% were ≥10 cm). The incidence remained relatively stable from 2004 to 2015, with an average of 150 cases per year.

Figure 1.

Flow diagram of case inclusion/exclusion. (ICD-O-3, International Classification of Diseases for Oncology, 3rd ed.)

Table 1.

Patient Cohort Characteristics

Characteristic	n	%
Year of diagnosis
2004	128	6.9
2005	146	7.9
2006	138	7.5
2007	148	8.0
2008	146	7.9
2009	161	8.7
2010	149	8.1
2011	172	9.3
2012	144	7.8
2013	130	7.0
2014	165	8.9
2015	173	9.4
2016	49	2.7
Sex
Male	935	50.6
Female	914	49.4
Age, y
18–49	540	29.2
50–59	458	24.8
60–69	446	24.1
≥70	405	21.9
Race/Ethnicity
White	1145	62.8
Black	399	21.9
Hispanic	100	5.5
Asian	149	8.2
Other	29	1.6
Insurance
Uninsured	88	4.9
Private	983	54.2
Medicaid	183	10.1
Medicare	559	30.8
Income
Below median	729	39.7
Above median	1106	60.3
Metropolitan area
Nonmetro	222	12.4
Metro	1575	87.7
Hospital type
Nonacademic	861	53.4
Academic	753	46.7
Charlson-Deyo score
0	1449	78.4
1	302	16.3
2	68	3.7
≥3	30	1.6
Masaoka staging
III	1108	59.9
IV	741	40.1
Metastatic disease
No	1266	68.5
Yes	583	31.5
Histologic subtype
A	132	7.1
AB	210	11.4
B1	222	12.0
B2	313	16.9
B3	393	21.3
NOS	579	31.3
Tumor size, cm
<5	233	15.3
5–9	846	55.7
10–14	346	22.8
15–19	62	4.1
≥20	32	2.1

NOS, not otherwise specified.

Treatment Modalities

Among the entire cohort, the most common mode of treatment was surgery plus radiation (25.8%), followed by definitive chemotherapy and/or radiation (23.7%) (Table 2). After stratifying by stage, the majority of patients with Masaoka stage III underwent surgical resection as part of their treatment (83.8%). The most common specific modalities were surgery plus radiation (32.6%), surgery alone (24.6%), surgery plus chemoradiation (16.7%), followed by surgery plus chemotherapy (9.9%). Stage IV patients also frequently had surgical resection as part of their treatment (60.2% total), but the most frequent specific modality was definitive chemotherapy and/or radiation (36.4%).

Table 2.

Frequency and Survival of Different Treatment Modalities for Stage III/IV Thymoma

Overall
Treatment	n	%	5-year Survival, %	95% CI
Chemotherapy/radiation	417	23.7	45.1	39.5, 50.6
None	32	1.8	38.1	16.8, 59.3
Surgery + chemo	203	11.5	70.1	61.7, 76.9
Surgery + chemoradiation	311	17.7	80.3	74.5, 85.0
Surgery + radiation	454	25.8	80.8	76.3, 84.5
Surgery only	344	19.5	71.5	65.6, 76.7
Stage III
Treatment	n	%	5-year Survival, %	95% CI
Chemotherapy/radiation	165	15.5	46.4	37.5, 54.8
None	8	0.8	73.3	28.7, 92.6
Surgery + chemo	106	9.9	67.6	56.3, 76.6
Surgery + chemoradiation	178	16.7	82.0	74.6, 87.5
Surgery + radiation	348	32.6	92.0	76.9, 86.1
Surgery only	263	24.6	73.9	67.1, 79.5
Stage IV
Treatment	n	%	5-year Survival, %	95% CI
Chemotherapy/radiation	252	36.4	44.2	36.9, 51.2
None	24	3.5	26.3	7.0, 51.1
Surgery + chemo	97	14.0	74.2	61.1, 83.4
Surgery + chemoradiation	133	19.2	77.8	67.5, 85.2
Surgery + radiation	106	15.3	76.6	65.4, 84.5
Surgery only	81	11.7	63.8	50.2, 74.6

CI, confidence interval.

Overall Survival

Overall 5-year survival was 70.3% (95% confidence interval [CI] 67.5%–72.9%) for stage III and 58.5% (95% CI 54.6%–62.2%) for stage IV thymoma. On Kaplan-Meier analysis, there were significant differences in survival based on treatment approach, with patients receiving no surgical treatment having the lowest survival in the overall cohort (45.1%, 95% CI: 39.5%–50.6%, Figure 2, log-rank P < .01). Among stage III patients, the highest 5-year survival was seen with treatments including surgical resection and radiation, specifically, surgery plus radiation (92.0%, 95% CI 76.9%–86.1%) and surgery plus chemoradiation (82.0%, 95% CI 74.6%–87.5%). Similarly, the best survival for stage IV patients was seen among patients treated with surgery plus chemoradiation (77.8%, 95% CI 67.5%–85.2%) and surgery plus radiation (76.6%, 95% CI 65.4%–84.5%). Patients treated with a definitive nonsurgical approach (chemotherapy and/or radiation) had a 5-year survival of 44.2% (95% CI 36.9%–51.2%).

Figure 2.

Survival of stage III/IV thymoma by treatment modality.

On adjusted analysis, treatment including surgery was substantially associated with improved survival compared with definitive chemotherapy and/or radiation, independent of other adjuvant or neoadjuvant treatments (Table 3). Other factors associated with mortality included increasing patient age, larger tumor size, presence of metastases, and treatment at a nonacademic institution. In the subset of surgically treated patients, there were no differences in mortality between surgery alone and any multimodality treatment including surgery on adjusted analysis (Table 4). However, complete (R0) resection was strongly associated with survival (hazard ratio 0.55, 95% CI 0.41–0.74) compared with incomplete (R1/R2) resection.

Table 3.

Adjusted Analysis of Factors Associated With Mortality Among Patients With Stage III/IV Thymoma

Characteristic	Hazard Ratio	95% Confidence Interval
Age, y	1.03	1.02	1.05
Sex
Male	Ref
Female	0.89	0.72	1.10
Race/Ethnicity
White	Ref
Black	1.18	0.89	1.56
Hispanic	1.25	0.72	2.19
Asian	1.24	0.84	1.83
Other	0.34	0.08	1.39
Insurance
Uninsured	0.78	0.40	1.51
Private	0.78	0.58	1.05
Medicaid	1.19	0.77	1.83
Medicare	Ref
Income
Below	Ref
Above	1.11	0.88	1.40
Metropolitan area
Nonmetro	Ref
Metro	1.15	0.83	1.59
Hospital type
Nonacademic	Ref
Academic	0.75	0.60	0.93
Charlson Score
0	Ref
1	1.14	0.88	1.49
2	1.23	0.78	1.92
≥3	1.52	0.70	3.27
Tumor size, cm
<5	Ref
5–9	1.30	0.94	1.80
10–14	1.55	1.07	2.25
15–19	2.12	1.24	3.64
≥20	2.20	1.12	4.31
Masaoka stage
III	Ref
IV	1.02	0.77	1.35
Histologic subtype
A	Ref
AB	1.01	0.60	1.69
B1	1.33	0.80	2.20
B2	1.52	0.94	2.45
B3	1.55	0.98	2.44
NOS	1.67	1.09	2.57
Metastases
No	Ref
Yes	1.47	1.09	1.99
Treatment
None	1.88	0.92	3.85
Chemotherapy/radiation	Ref
Surgery + chemotherapy	0.65	0.45	0.93
Surgery + chemoradiation	0.37	0.26	0.54
Surgery + radiation	0.41	0.30	0.55
Surgery only	0.51	0.37	0.70

Table 4.

Adjusted Analysis of Factors Associated With Mortality Among Surgically Treated Patients With Stage III/IV Thymoma

Characteristic	Hazard Ratio	95% ConfidenceInterval
Age, y	1.04	1.02	1.06
Sex
Male	Ref
Female	0.80	0.60	1.06
Race/Ethnicity
White	Ref
Black	1.39	0.94	2.04
Hispanic	1.13	0.55	2.32
Asian	1.36	0.82	2.23
Other	.	.	.
Insurance
Uninsured	0.65	0.28	1.48
Private	0.62	0.42	0.90
Medicaid	0.90	0.50	1.60
Medicare	Ref
Income
Below	Ref
Above	1.04	0.76	1.43
Metropolitan area
Nonmetro	Ref
Metro	0.96	0.63	1.46
Hospital type
Non-academic	Ref
Academic	0.78	0.58	1.04
Charlson
0	Ref
1	1.37	0.98	1.91
2	1.21	0.71	2.07
≥3	0.62	0.15	2.59
Tumor size, cm
<5	Ref
5–9	1.26	0.83	1.91
10–14	2.14	1.32	3.48
15–19	2.96	1.43	6.12
≥20	2.15	0.85	5.40
Masaoka stage
III	Ref
IV	1.07	0.76	1.50
Histologic subtype
A	Ref
AB	1.09	0.52	2.27
B1	1.83	0.88	3.79
B2	2.08	1.05	4.10
B3	1.70	0.88	3.29
NOS	1.80	0.93	3.49
Metastases
No	Ref
Yes	1.16	0.79	1.73
Completeness of resection
R0	Ref
R1/R2	0.55	0.41	0.74
Treatment
Surgery only	Ref
Surgery + chemotherapy	1.25	0.81	1.93
Surgery + chemoradiation	0.67	0.42	1.06
Surgery + radiation	0.81	0.57	1.15

NOS, not otherwise specified.

Neoadjuvant and Adjuvant Therapies

In the subset of patients with nonmetastatic disease (n = 1266), 1013 underwent surgical resection. The analysis of overall survival was repeated in this cohort, categorizing patients based on presence or absence of any neoadjuvant and/or adjuvant therapy in addition to surgery (Appendix 2). Overall R0 resection rate in this nonmetastatic cohort was 54.7%. In this cohort, 160 (15.8%) patients received some form of neoadjuvant therapy (0.9% radiation, 12.6% chemotherapy, 2.3% chemoradiation). All types of neoadjuvant therapy were combined into 1 cohort for analysis given the limited number of patients. Complete (R0) resection was achieved in 57.2% of patients receiving any induction therapy, compared with 54.2% of those treated with primary surgical resection (P = .50). On adjusted logistic regression, lower odds of R0 resection were associated with more aggressive histologic subtypes and Masaoka stage IV (odds ratio 0.64, 95% CI 0.41–0.99), while induction therapy was not significantly associated with R0 resection (odds ratio 1.20, 95% CI 0.75–1.93) (Appendix 3).

In the same cohort of nonmetastatic patients, 47.6% of those having R0 resection had adjuvant radiation compared with 52.4% with incomplete (R1/R2) resection receiving adjuvant radiation (P < .01). An analysis of overall survival, stratified by completeness of resection, showed that adjuvant radiation was significantly associated with longer survival for patients who had incomplete resections (log-rank P < .01), but not for those with a complete R0 resection (log-rank P = .20) (Figure 3).

Figure 3.

Survival for surgery only vs adjuvant radiation for stage III/IV nonmetastatic thymoma, stratified by completeness of resection.

Comment

Optimal treatment strategies for advanced stage thymoma are unclear, with no level-1 evidence to guide care. The rarity of the disease precludes well-powered prospective trials, making it a good disease process to investigate using a large database approach. In a retrospective review of a national database from 2004 to 2016, we found significant variation in treatment approaches to stage III and IV thymoma. Long-term overall survival differed markedly between treatment modalities.

It is well established that surgical resection is the mainstay of thymoma treatment, and can improve survival even in patients with stage IV disease.^15,16 We found that this was reflected in real-world practice, with treatment approaches including surgical resection being the most common for both stage III and IV thymoma in the United States. Surgical resection was often one component of multimodality therapy. Nonetheless, it is important to note that 16.3% of patients with stage III disease did not undergo resection. Similarly, the majority of stage IV patients also had a resection (60.2%), but more than one third were treated with definitive chemotherapy and/or radiation.

Overall 5-year survival was 70.3% for stage III and 58.5% for stage IV, comparable to 60%–70% reported in the literature for combined stage III/IV.^17,18 The best survival for both stages was seen when surgical resection was combined with radiation therapy. This is consistent with a number of previous studies, including large retrospective databases and small prospective series, which have demonstrated improved survival for surgical resection followed by adjuvant radiation for stage IIb to IV thymoma.^7,9–11 Although there is some debate in the literature regarding adjuvant radiation for stage IIb, there is general consensus that adjuvant radiation is beneficial in stage III and IV resected disease. Many of these studies did not investigate important covariates, most notably completeness of resection. When stratifying by completeness of resection, we found that the survival benefit conferred by radiation treatment did not persist among those patients who had a complete (R0) resection. Although some prior work has suggested this dichotomy,⁷ some of these studies included thymic carcinoma, which we notably excluded, and for which radiation may be more beneficial than for thymoma.^9,10

The role of neoadjuvant treatment for advanced stage disease has been the subject of some small prospective series. In a phase II study of neoadjuvant chemoradiation, with 22 patients with locally advanced thymoma or thymic carcinoma treated at 4 institutions, Korst and associates¹² reported a 45% partial radiographic response and 77% R0 resection rate.¹² In a retrospective study of 56 patients treated for stage III or IVA thymic tumors at a single institution, Lucchi and coworkers¹³ reported that patients treated with neoadjuvant chemotherapy, surgery, and adjuvant radiation had improved R0 resection rate and survival compared with those treated with primary surgery and adjuvant radiation.¹³ These studies and others suggest that there is an oncologic and survival benefit associated with induction therapy, which has been attributed to increasing the likelihood of achieving a complete resection.¹⁹ In a subanalysis of patients with nonmetastatic stage III and IV disease, we found no association between induction therapy (including any of chemotherapy and/or radiation therapy) and completeness of resection. There was also no significant difference in survival between these groups. This suggests that induction therapy may not yet be a generalizable approach to improve R0 resection or survival outside of the clinical trial setting. Given the small proportion of patients treated with neoadjuvant therapy in general, we are unable to comment on specific modalities or differences between neoadjuvant radiation, chemotherapy, or combinations. We do note that neoadjuvant radiation was rarely used, with neoadjuvant chemotherapy being much more common.

Our findings should be interpreted in light of several limitations. First, this was a retrospective investigation and is prone to the potential for selection bias and unmeasured confounding. For example, healthier patients or those with less-advanced disease may have been preferentially selected for certain therapies (ie, surgical resection), potentially contributing to perceived improvements in survival not directly attributable to the treatment itself. Selection bias could also impact our interpretation of the findings regarding R0 resection and induction therapy. Patients with unresectable disease are likely to be selected to have induction therapy. If these patients are subsequently able to undergo resection, having an R0 resection rate equivalent to those receiving upfront surgery may actually suggest a benefit for these previously unresectable patients. Second, the NCDB does not include information regarding recurrence of disease or disease-specific survival. In a relatively indolent and slowly progressing disease such as thymoma, these are critically important outcomes to investigate in addition to overall survival. Third, in our investigation of induction and adjuvant therapies, we were not able to analyze the impact of potential differences between treatment with specific chemotherapeutic agents or doses of radiation. Thus, the induction therapy cohort likely represents a heterogenous group of treatment approaches. There is also the possibility that some patients treated with induction were downstaged to Masaoka I/II and therefore not included in our cohort. Finally, specific classification systems such as the TNM staging criteria are not available for thymoma in the NCDB. Therefore, we had to rely on the tumor extension variable to approximate Masaoka staging. Although this limits our ability to distinguish between different subgroups of advanced thymoma, it does sufficiently define a cohort with tumor extension beyond the gland.

There is limited evidence guiding the care of patients with advanced stage thymomas, reflected in a multitude of treatments offered. When possible, surgical resection followed by adjuvant radiation is associated with the longest survival for both stage III and IV. Completeness of resection remains the most important prognostic factor, and adjuvant radiation is especially important for those with incomplete resections. Induction therapy does not appear to be associated with completeness of resection in this national retrospective cohort, but further prospective studies are needed to appropriately evaluate its role in the treatment of advanced stage thymoma.