Prognostic factors in thyroid cancer with lung-only metastasis: SEER-based retrospective study Authors: Wei-Hu Xia, Yan Su, Qin-Yang Wang, Wei Wan, Ya-Min Zhang, Jin-Peng Liu Published: 2026 ABSTRACT BackgroundLung-only metastasis in thyroid cancer is uncommon, and its prognostic determinants remain poorly defined. This study aimed to investigate the clinicopathological features and prognostic factors associated with survival in patients with primary thyroid cancer with lung-only metastases (PTCSLMs).MethodsPatients diagnosed with PTCSLMs from 2010 to 2015 were screened from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) database. Inclusion criteria comprised histologically confirmed thyroid cancer, first primary malignancy, and lung‑only metastasis at initial diagnosis, while patients with bone/liver/brain metastases, unknown race, unknown T stage/T0, or missing survival data were excluded. Variables analyzed included demographics (age, sex, marital status, race), clinicopathological features (histological subtype, T stage, N stage, grade stage), and treatment modalities {surgery, radiotherapy [external beam radiation therapy (EBRT)/radioactive iodine (RAI)/others], and chemotherapy}. Overall survival (OS) was defined as time from diagnosis to death from any cause, and cancer-specific survival (CSS) as time to thyroid cancer‑specific death. Univariate and multivariate Cox proportional hazards model were used to identify prognostic factors of PTCSLMs. The Kaplan-Meier method was used to generate survival curves, and the difference was compared using the log-rank test.ResultsA total of 592 patients diagnosed with PTCSLMs from 2010 to 2015 were extracted from the SEER database based on the specified criteria for inclusion and exclusion. Univariate analysis demonstrated statistically significant associations between OS and age at diagnosis, marital status, histologic subtype, T stage, N stage, grade stage, surgery, radiotherapy and chemotherapy (P<0.001). The results of multivariate survival analyses showed that age at diagnosis [43–80 vs. 6–42 years, hazard ratio (HR) =5.044, 95% confidence interval (CI): 2.290–11.111, P<0.001; 81–99 vs. 6–42 years, HR =7.769, 95% CI: 3.296–18.312, P<0.001], histologic subtype [others vs. papillary adenocarcinoma (PA), HR =1.786, 95% CI: 1.323–2.411, P<0.001], T stage (T2 vs. TX, HR =0.480, 95% CI: 0.247–0.935, P=0.03), grade stage (poorly differentiated vs. well differentiated, HR =1.976, 95% CI: 1.123–3.476, P=0.02; undifferentiated vs. well differentiated, HR =2.531, 95% CI: 1.417–4.521, P=0.002), surgery (total thyroidectomy vs. refuse/none, HR =0.387, 95% CI: 0.278–0.538, P<0.001; others vs. refuse/none, HR =0.663, 95% CI: 0.455–0.967, P=0.03), radiotherapy (EBRT vs. refuse/none/unknown, HR =0.734, 95% CI: 0.564–0.955, P=0.02; RAI vs. refuse/none/unknown, HR =0.481, 95% CI: 0.336–0.688, P<0.001; others vs. refuse/none/unknown, HR =0.498, 95% CI: 0.254–0.978, P=0.043) and chemotherapy (yes vs. no/unknown, HR =0.741, 95%CI: 0.560–0.982, P=0.04) were independent prognostic factors of PTCSLMs.ConclusionsOur study suggested that the higher age at diagnosis and higher pathological grade were correlated with worse prognosis. In contrast, surgery, radiotherapy and chemotherapy showed a favorable prognostic association. Age at diagnosis, histological type, T stage, grade stage, surgery, radiotherapy and chemotherapy are associated with prognosis and they are independent prognostic factors for PTCSLMs. Highlight box Key findings • Age at diagnosis, histological type, T stage, grade stage, surgery, radiotherapy and chemotherapy are associated with prognosis and are independent prognostic factors for primary thyroid cancer with lung-only metastases (PTCSLMs). What is known and what is new? • The incidence of lung metastasis in thyroid carcinoma is generally considered low, whereas PTCSLMs is extremely rare. In previous studies, risk factors and outcomes of PTCSLMs have been examined, but no unified conclusion has been reached. • This is the first comprehensive study to identify risk factors (include surgical procedure, radiotherapy, and chemotherapy) associated with PTCSLMs. What is the implication, and what should change now? • By identifying relevant risk factors, this study aims to provide personalized diagnostic and therapeutic strategies for these patients. Introduction In recent years, thyroid carcinoma (TC) has increased in incidence worldwide, raising concerns in the academic community about this most common endocrine tumor (1). In general, TC exhibits an extensive range of clinical behavior, from indolent cancers with high survival rates to highly aggressive cancers. There is also considerable variability in the prognosis of patients with TC (2,3). As the metastatic rate of thyroid cancer to distant sites is only 2%, the outcomes worsen with distant metastases, which accounts for 73.2% of thyroid cancer-specific deaths (4). Patients with TC are most likely to develop metastatic lung disease, followed by bone, liver, and brain disease (5-7). Metastases of thyroid cancer are usually asymptomatic and difficult to distinguish until they are detected by a full-body metastatic examination. Lung metastases are typically overlooked at the initial diagnosis of TC due to their asymptomatic nature. In comparison to patients without distant metastases, survival for TC patients with lung metastases decreases by approximately 50% after 20 years (51.2% vs. 98%) (8). The incidence of lung metastasis in TC is generally considered low, whereas primary thyroid cancer with lung-only metastasis (PTCSLM) is extremely rare. It was reported that about 7–30% of differentiated TC childhood patients suffered from lung metastasis, while that in the adults was less than 4% (9,10). For TC patients with lymph node metastasis (LNM) or distant metastases, radioactive iodine (RAI) 131 therapy is the main treatment strategy (11,12). Patients who are resistant to RAI 131 are found to have significantly reduced survival rates. The treatment of distant metastases has improved and developed in recent years. In spite of this, PTCSLMs still has poor survival rates due to late detection, making it the leading cause of deaths (13). According to a retrospective study, identifying asymptomatic distant metastases early is beneficial to long-term survival in TC patients (14). In order to improve the prognosis of TC patients, it is of great importance to identify patients with a high risk of lung-only metastasis. While isolated lung metastasis is relatively uncommon in thyroid cancer, its clinical significance extends far beyond its low prevalence (15). Unlike multiorgan or extra‑pulmonary dissemination, which typically portends rapid progression and limited therapeutic options, isolated pulmonary involvement often follows a more protracted course and allows for locoregional interventions (16,17). Furthermore, emerging genomic data indicate that lung metastases may harbor distinct oncogenic drivers, raising the possibility of targeted approaches tailored to this subset (18). Therefore, we reasoned that separating this cohort from other patterns of distant spread would not only reduce clinical heterogeneity but also uncover prognostic and therapeutic insights that could inform personalized management (15). In previous studies, risk factors and outcomes of PTCSLMs have been examined, but no unified conclusion has been reached. Many scholars (19-21) have analyzed the risk factors and outcomes of PTCSLMs. However, they only included the demographic and clinicopathological features of PTCSLMs. Using the Surveillance, Epidemiology, and End Results (SEER) database, in addition to demographic and clinicopathological features, we also included treatment modalities, including surgical procedure, radiotherapy, and chemotherapy, in our study, aimed to identify independent prognostic factors and analyze the overall survival (OS) for PTCSLMs. To our knowledge, our study is the first comprehensive study to identify risk factors associated with PTCSLMs. The purpose of this study is to provide some basis for PTCSLMs screening and preventative interventions for TC patients. We present this article in accordance with the STROBE reporting checklist (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0333/rc). Methods Data sources This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. We conducted a retrospective cohort study using the linked SEER registries, which cover approximately 28% of the U.S. population and contain approximately 97% of all cancer cases (22,23). Relevant clinical information was obtained from the “incidence - SEER Research Plus data, 18 Registries, Nov 2020 Sub (2000–2018)” using SEER*Stat 8.3.9.2 software for the period 2010 to 2015. No formal sample‑size calculation was performed. Because this was a retrospective analysis of the SEER database, we included all patients who met the predefined eligibility criteria during the study period. Thus, the final sample size was determined by the number of eligible cases available in the database. Inclusion and exclusion criteria Inclusion criteria: (I) confirmed the diagnosis of thyroid cancer; (II) the first indicator of malignancy; (III) lung-only metastasis at the time of diagnosis; (IV) diagnosis time: 2010–2015. Exclusion criteria: (I) with bone, liver, and brain metastasis; (II) unknown race; (III) unknown T stage and T0; (IV) unknown survival time. Study variables and outcomes We filtered the patients’ demographic characteristics, including gender (female and male), ethnicity (White, Black and others ethnicity), age of diagnosis, marital status (single, married, divorced, separated, widowed, and unknown), clinicopathological features, including histological type [papillary adenocarcinoma (PA), papillary with follicular variant (PFV), follicular adenocarcinoma (FA), undifferentiated carcinoma (UC), and others type], T stage (TX, T1–4), N stage (NX, N0 and N1), and grade stage (well differentiated, moderately differentiated, poorly differentiated, undifferentiated, and unknown), treatment modalities, including surgical procedure (refuse/none, total thyroidectomy, others surgical procedure, and unknown), radiotherapy [refuse/none/unknown, RAI, external beam radiation therapy (EBRT), and others radiation therapy methods], chemotherapy (no/unknown and yes), and prognosis status such as survival status (dead and alive) and survival time, etc. The others histological subtypes included “acinar cell carcinoma”, “sclerosing carcinoma”, “insular carcinoma”, “oxyphilic adenocarcinoma”, “adenocarcinoma”, “squamous cell carcinoma”, “carcinoma with osteoclast-like giant cells”, “pseudosarcomatous carcinoma”, “spindle cell carcinoma”, “giant cell and spindle cell carcinoma”, “anaplastic carcinoma”, and “carcinoma, NOS”. Other surgical procedures included local tumor destruction and partial thyroidectomy. The end-point of our study was OS and cancer-specific survival (CSS). OS was calculated from the time of confirmed the diagnosis of cancer to the date of patient death from any cause (24). CSS referred to the duration between the diagnosis of cancer and death specifically linked to the primary tumor or the most recent follow-up date (25). Statistical analysis Age of diagnosis cutoff values were determined using software called X-tile. We chose X‑tile because it provides a data‑driven, outcome‑based approach to identify optimal cut‑off values for continuous variables, specifically using the maximal Chi-squared method (or log‑rank test statistics) to stratify patients into clinically meaningful prognostic groups (26). This approach has been widely adopted in SEER‑based and oncological studies, as it offers a reproducible and transparent way to define cut‑points that are directly relevant to survival outcomes (26). Moreover, by using X‑tile, we aimed to enhance clinical interpretability—age groups defined by prognostic impact are more intuitive for clinicians than a per‑year hazard ratio (HR), especially when designing risk‑stratified treatment strategies. Age of diagnosis was divided into three groups (6–42, 43–80, and 81–99 years) according to the X-tile software. First, we conducted a descriptive analysis to describe the characteristics of the populations. Then, we performed univariate Cox regression analysis. Additionally, an analysis of multivariate Cox was performed with factors with a P value below 0.05 in the univariate Cox analysis. Each factor was estimated by calculating its HR and 95% confidence interval (95% CI). Finally, by using the Kaplan-Meier method, survival curves were drawn and used to calculate 3- and 5-years OS rates. A log-rank test was performed on survival curves to determine whether there was a significant difference between them. Consider to TX was not treated merely as a “missing” value but rather represented a distinct clinical subgroup with specific characteristics. Interestingly, preliminary Kaplan-Meier curves indicated that this group had the worst prognosis compared to the measurable T-stages. Therefore, we deliberately set TX as the reference baseline to allow for a direct comparison of the HRs for all other valid T-stages against this specific subgroup. Another reason for choosing TX as the reference was to ensure statistical stability, as this category contained the intermediate number of patients, providing the most robust baseline variance for the multivariate model. The level of P<0.05 was considered to the threshold for statistical significance, and that all reported confidence intervals were calculated at the 95% level using two-sided tests. All data analysis was conducted using SPSS version 25.0 software. Results Age of diagnosis grouped by X-tile To determine the best cutoff value of age of diagnosis, X-tile software was used for grouping in our study. The age of diagnosis was divided into three groups: 6–42, 43–80, and 81–99 years (Figure 1). The best cut-off age group of PTCSLMs according to the X-tile software. PTCSLM, primary thyroid cancer with single lung metastasis. Patient demographic and clinical data A total of 592 patients diagnosed with PTCSLMs from 2010 to 2015 were extracted from the SEER database based on the specified criteria for inclusion and exclusion. In Table 1, we present the clinical characteristics of the study population. Median age at diagnosis was 67 years, majority of the patients (72.0%) were ranged from 43 to 80 years, and male: female ratio was approximately 1.7:2. Most of them (77.4%) were white. In histological type, they consisted of 230 (38.9%) papillary adenocarcinoma, 74 (12.5%) PFV carcinoma, 55 (9.3%) FA, 3 (0.5%) UC, and 230 (38.9%) others uncommon pathological type. Among the cohort, the majority presented with advanced stage. T4 stage was the most common, accounting for 329 cases (55.6%), followed by T3 stage with 135 cases (22.8%). T2 stage and T1 stage were less frequent, with 33 (5.6%) and 29 (4.9%) cases, respectively. TX stage (stage unknown) was recorded in 66 cases (11.1%). For nodal status, N1 stage was the dominant category, observed in 373 patients (63.0%). N0 stage (node‑negative) accounted for 162 cases (27.4%), while NX stage (nodal status not assessed) was reported in 57 cases (9.6%). 239 (40.4%) was not underwent surgery. Among these patients who underwent surgery, 292 (49.3%) underwent total thyroidectomy, 52 (8.8%) underwent others surgery type, and 9 (1.5%) of surgery type was unknown. Besides, 122 (20.6%) and 198 (33.4%) underwent EBRT and RAI, respectively. At the same time, only a few patients (n=110, 18.6%) underwent chemotherapy. Table 1Baseline demographic and clinicopathologic characteristics for PTCSLMsCharacteristicsValueAge at diagnosed (years)67 [6–99] 6–4269 (11.7) 43–80426 (72.0) 81–9997 (16.4)Year of diagnosis 2010108 (18.2) 201183 (14.0) 201286 (14.5) 2013105 (17.7) 2014104 (17.6) 2015106 (17.9)Sex Female319 (53.9) Male273 (46.1)Marital status Single132 (22.3) Married295 (49.8) DSW146 (24.7) Unknown19 (3.2)Race Black48 (8.1) White458 (77.4) Others86 (14.5)Histological type PA230 (38.9) PFV74 (12.5) FA55 (9.3) UC3 (0.5) Others230 (38.9)T stage TX66 (11.1) T129 (4.9) T233 (5.6) T3135 (22.8) T4329 (55.6)N stage NX57 (9.6) N0162 (27.4) N1373 (63.0)Grade stage Well differentiated47 (7.9) Moderately differentiated25 (4.2) Poorly differentiated55 (9.3) Undifferentiated177 (29.9) Unknown288 (48.6)Surgery Refuse/none239 (40.4) Total thyroidectomy292 (49.3) Others52 (8.8) Unknown9 (1.5)Radiotherapy Refuse/none/unknown250 (42.2) EBRT122 (20.6) RAI198 (33.4) Others22 (3.7)Chemotherapy No/unknown482 (81.4) Yes110 (18.6)Survival status Dead392 (66.2) Alive200 (33.8)Survival time (month)13.5 [0–107]Data are presented as median [range] or n (%). DSW, divorced + separated + widowed; EBRT, external beam radiation therapy; FA, follicular adenocarcinoma; N, node; PA, papillary adenocarcinoma; PFV, papillary with follicular variant; PTCSLM, primary thyroid cancer with single lung metastasis; RAI, radioactive iodine; T, tumor; UC, undifferentiated carcinoma. Baseline demographic and clinicopathologic characteristics for PTCSLMs Data are presented as median [range] or n (%). DSW, divorced + separated + widowed; EBRT, external beam radiation therapy; FA, follicular adenocarcinoma; N, node; PA, papillary adenocarcinoma; PFV, papillary with follicular variant; PTCSLM, primary thyroid cancer with single lung metastasis; RAI, radioactive iodine; T, tumor; UC, undifferentiated carcinoma. Risk factors for patients with PTCSLMs The univariate analysis demonstrated that age at diagnosed (P<0.001), marital status (P<0.001), histological type (P<0.001), T stage (P<0.001), N stage (P<0.001), grade stage (P<0.001), surgery (P<0.001), radiotherapy (P<0.001), and chemotherapy (P<0.001) were associated with prognosis for PTCSLMs patients. However, multivariate Cox regression analysis showed independent prognostic factors only include age at diagnosed (43–80 vs. 6–42 years, HR =5.044, 95% CI: 2.290–11.111, P<0.001; 81–99 vs. 6–42 years, HR =7.769, 95% CI: 3.296–18.312, P<0.001), histological type (others vs. PA, HR =1.786, 95% CI: 1.323–2.411, P<0.001), T stage (T2 vs. TX, HR =0.480, 95% CI: 0.247–0.935, P=0.03), grade stage (poorly differentiated vs. well differentiated, HR =1.976, 95% CI: 1.123–3.476, P=0.02; undifferentiated vs. well differentiated, HR =2.531, 95% CI: 1.417–4.521, P=0.002), surgery (total thyroidectomy vs. refuse/none, HR =0.387, 95% CI: 0.278–0.538, P<0.001; others vs. refuse/none, HR =0.663, 95% CI: 0.455–0.967, P=0.03), radiotherapy (EBRT vs. refuse/none/unknown, HR =0.734, 95% CI: 0.564–0.955, P=0.02; RAI vs. refuse/none/unknown, HR =0.481, 95% CI: 0.336–0.688, P<0.001; Others vs. refuse/none/unknown, HR =0.498, 95% CI: 0.254–0.978, P=0.043), and chemotherapy (yes vs. no/unknown, HR =0.741, 95% CI: 0.560–0.982, P=0.04) (Table 2). Table 2Univariate and multivariable analysis for PTCSLMsCharacteristicsUnivariate analysisMultivariate analysisHR95% CIPHR95% CIPSex FemaleRef Male0.9730.798–1.1870.79Age at diagnosed (years) 6–42RefRef 43–8010.8065.101–22.890<0.001***5.0442.290–11.111<0.001*** 81–9921.83910.074–47.341<0.001***7.7693.296–18.312<0.001***Marital status SingleRefRef Married1.4611.113–1.9180.006**1.1150.831–1.4980.47 DSW1.8461.370–2.487<0.001***1.1540.814–1.6380.42 Unknown0.8990.476–1.6980.740.8000.412–1.5530.51Race BlackRef White0.9410.647–1.3700.75 Others1.0090.649–1.5680.97Histological type PARefRef PFV0.6500.430–0.9830.041*0.8530.555–1.3100.47 FA1.1950.812–1.7590.371.2220.809–1.8440.34 UC4.1051.303–12.9320.02*1.1440.344–3.8090.83 Others4.3513.438–5.506<0.001***1.7861.323–2.411<0.001***T stage TXRefRef T10.1760.084–0.371<0.001***0.6270.283–1.3890.25 T20.2390.127–0.447<0.001***0.4800.247–0.9350.03* T30.2290.155–0.338<0.001***0.7470.470–1.1870.22 T41.0110.755–1.3550.941.3590.916–2.0150.13N stage NXRefRef N00.4830.341–0.683<0.001***0.7280.500–1.0590.10 N10.5440.399–0.743<0.001***1.0370.731–1.4720.84Grade stage Well differentiatedRefRef Moderately differentiated0.8690.393–1.9210.730.9290.408–2.1150.86 Poorly differentiated3.4211.994–5.867<0.001***1.9761.123–3.4760.02* Undifferentiated9.6675.918–15.792<0.001***2.5311.417–4.5210.002** Unknown1.4600.905–2.3550.120.9210.558–1.5190.75Surgery Refuse/noneRefRef Total thyroidectomy0.1480.117–0.188<0.001***0.3870.278–0.538<0.001*** Others0.4070.289–0.572<0.001***0.6630.455–0.9670.03* Unknown0.2950.131–0.6660.003**0.4570.198–1.0560.07Radiotherapy Refuse/none/unknownRefRef EBRT1.1900.942–1.5020.140.7340.564–0.9550.02* RAI0.1840.139–0.244<0.001***0.4810.336–0.688<0.001*** Others0.2730.145–0.516<0.001***0.4980.254–0.9780.043*Chemotherapy No/unknownRefRef Yes1.9131.513–2.418<0.001***0.7410.560–0.9820.04**, P<0.05; **, P<0.01; ***, P<0.001. CI, confidence interval; DSW, divorced + separated + widowed; EBRT, external beam radiation therapy; FA, follicular adenocarcinoma; HR, hazard ratio; N, node; PA, papillary adenocarcinoma; PFV, papillary with follicular variant; PTCSLM, primary thyroid cancer with single lung metastasis; RAI, radioactive iodine; T, tumor; UC, undifferentiated carcinoma. Univariate and multivariable analysis for PTCSLMs *, P<0.05; **, P<0.01; ***, P<0.001. CI, confidence interval; DSW, divorced + separated + widowed; EBRT, external beam radiation therapy; FA, follicular adenocarcinoma; HR, hazard ratio; N, node; PA, papillary adenocarcinoma; PFV, papillary with follicular variant; PTCSLM, primary thyroid cancer with single lung metastasis; RAI, radioactive iodine; T, tumor; UC, undifferentiated carcinoma. Survival outcomes In total, 392 (66.2%) PTCSLMs patients died and median survival time was 13.5 months in follow‐up. The longest survival time was 107 months. The 3-year OS rate was 42.6%, and CSS rate was 47.3% (Figure 2A,2B). The 5-year OS rate was 35.9%, and CSS rate was 40.7% (Figure 2A,2B). Figure 3A-3D shows the survival curves stratified by age at diagnosis, surgery, radiotherapy and chemotherapy. When stratified by age, 3- and 5-year OS were 92.5% vs. 89.1%, 40.2% vs. 33.0%, and 18.6% vs. 12.4% for patients with 6–42, 43–80 and 81–99 years, respectively (Figure 3A). When stratified by surgery, 3- and 5-year OS were 8.8% vs. 6.3%, 71.6% vs. 62.1%, 34.6% vs. 24.4%, and 55.6% vs. 44.4% for patients with none surgery, total thyroidectomy, others surgery, and unknown, respectively (Figure 3B); 3- and 5-year OS were 21.6% vs. 19.7%, 11.2% vs. 7.4%, 85.0% vs. 71.6%, and 72.7% vs. 56.1% for patients with none radiotherapy, ERBT, RAI and others, respectively (Figure 3C). Finally, as shown in Figure 3D, the 3-year OS rates for patients with and without chemotherapy were 48.8% and 15.1%, respectively, and the corresponding 5-year OS rates were 41.5% and 11.1%, respectively. Kaplan-Meier curves for PTCSLMs: (A) OS; (B) CSS. CSS, cancer-specific survival; OS, overall survival; PTCSLM, primary thyroid cancer with single lung metastasis. OS Kaplan-Meier curves for PTCSLMs: (A) age; (B) surgery; (C) radiotherapy; (D) chemotherapy. OS, overall survival; PTCSLM, primary thyroid cancer with single lung metastasis. Discussion The main metastatic sites of TC are lung, bone, liver and brain, among which, the most common metastatic site is the lung, followed by bone (27). Nearly 50% of autopsies of patients who died of TC reveal lung metastasis. Lung metastasis of TC appears as multiple small nodular shadows on imaging examination but does not show typical symptoms (28). Only 15% to 20% of patients with metastatic lung carcinoma present with clinical symptoms such as cough or nonspecific chest pain (29). Cough, hemoptysis and dyspnea are the common signs of lung metastasis, which quickly worsen the quality of life and speed up the progression towards death. TC metastases only to the lung are generally thought to be rare. Therefore, there are few reports of related research on PTCSLMs patients. In our study, we came up with some interesting conclusions. We found seven independent predictors associated with PTCSLMs: age at diagnosis, histology type, T stage, grade stage, surgery, radiotherapy, and chemotherapy. Previous study has found that increasing age further influences the progression and occurrence of lung metastasis in TC patients (30). Several studies have performed survival analyses among differentiated TC patients with lung metastasis, which revealed that age, postoperative serum thyroglobulin level, cumulative RAI dose, and tumor size were important prognostic factors (31-33). A study performed by Sugino et al. (34) on a group of 106 follicular TC patients with distant metastasis has demonstrated that the 5-, 10-, and 15-year were 82.0%, 63.8% and 23.9%, respectively. And they also found tumor size and age at diagnosis are the most important prognostic factors for follicular TC patients with distant metastasis. Others studies (35,36) also found age exhibit a proportional relationship with the risk of death for undifferentiated TC patients. Similarly, we have reached similar conclusions: higher age at diagnosis was associated with the worse prognosis in our study (81–99 vs. 6–42 years, HR =7.769, 95% CI: 3.296–18.312, P<0.001; 43–80 vs. 6–42 years, HR =5.044, 95% CI: 2.290–11.111, P<0.001). First and most simply, it presented with insidious onset and atypical clinical signs and symptoms, and chest computed tomography (CT) showed multiple small nodules in early stage for PTCSLMs. Second, the higher the age at diagnosis, the later they may diagnosis, and they were more likely to be combined with others chronic disease and others metastatic sites such as lymph nodes, bone, and so on. Notably, lung metastasis was more common in pediatric TC, with an incidence of 19% according to Alzahrani et al. (10), but under 42 years of age accounted for only 69 (11.7%) patients in all PTCSLMs. There has been ample evidence that clinicopathological features play a significant role in the progression of cancer. In TC patients, LNM is associated with pulmonary metastatic carcinogenesis and closely related to lung metastasis (36,37). According to Liu et al. (20), age, T-stage, and N-stage as well as histological type were independent risk factors for PTCSLMs. In general, it was known that lateral LNM invaded distant organs more readily than central LNM. However, in our study, LNM did not predict survival in PTCSLMs, contrary to previous reported. The outcome could be attributed to several factors. Firstly, the study’s small sample size may explain this difference, which may reflect limited statistical power, heterogeneity among thyroid cancer subtypes, incomplete information on nodal burden, or the fact that distant metastatic disease already dominates prognosis in this selected population. Secondly, the lymph node ratio was defined as the number of metastatic lymph nodes divided by the number of lymph nodes removed, which was regarded as a measure of both LNM and completeness of lymph node removal, as well as the maximum diameter of the tumor in the metastatic lymph node. Many studies have demonstrated that high lymph node ratio and large lymph node metastases were significant risk factors for TC recurrence (38-40). Thus, the lymph node ratio, lymph node size, and number of metastatic lymph nodes influence thyroid cancer progression and prognosis. Due to the limitations of incomplete data records in the SEER database, we were unable to include the corresponding information in our study, leading to results that might be contrary to conventional knowledge. For this reason, future studies must include larger populations and prospectively rigorous designs. We also concluded that poor-differentiation TC (poorly differentiated and undifferentiated) were more susceptible to lung metastasis. The reason for this may be the fact that cancer cells have the capacity to invade surrounding tissues and capillaries, causing them to become more prone to developing early metastases. The results of our study are consistent with those of previous studies (41). The most common primary malignancy of the lungs is squamous cell carcinoma of the lungs, making pathological diagnosis of suspicious pulmonary lesions crucial in clinical diagnosis. By using PAX-8 immunohistochemical staining, Kao et al. (42) suggested that the nature of the pulmonary lesions could be further determined. Furthermore, hematologic metastasis is an important route for metastatic pulmonary carcinoma, and a study of deceased TC patients showed that follicular TC and anaplastic TC metastasize to the pulmonary lungs more frequently (7). Despite previous studies finding an increased risk of developing distant metastasis in various subtypes of TC associated with alterations in relevant molecular biological markers such as TP53 mutations, RET, TERT promoter, and BRAF, there are still many unknowns (43-46). According to American Thyroid Association guidelines published in 2009, patients with pulmonary metastases from differentiated TC should undergo total or subtotal thyroidectomy followed by RAI (47). The majority of studies on this treatment have shown significant efficacy (48,49). A meta-analysis results showed that the overall response rate of RAI was 58% (12). Although some studies (4,50,51) have shown low efficacy rates for RAI treatment of differentiated TC with lung metastasis, other studies have shown higher effectiveness (48,52-54). The differences may be due to the different criteria for selecting dosage and evaluating efficacy. Besides, the efficacy of RAI was associated with the expression of sodium iodide symporter. Finally, RAI efficacy of TC was directly related to impaired iodine uptake (55). Our study also suggested PTCSLMs patients treated with surgery, radiotherapy and chemotherapy were independently associated with better OS. Residual thyroid tissue after cervical thyroidectomy increased the risk of recurrence for TC patients. Radiation therapy in 131I was usually used to damage residual thyroid tissue in order to minimize cancer recurrence. However, RAI had a high incidence of adverse reactions, and some patients died as a result. In addition, patients with progressive locally advanced and RAI-resistant differentiated TC may benefit significantly from apatinib, lenvatinib, and sorafenib as part of a systematic therapy plan (56-58). In this study, we used data from the SEER database. The SEER database represents approximately 28% of the US population, which may represent better clinical outcomes for patients with TC who have lung metastasis than one-center studies. Our study has some notable limitations, as is inevitable. First, selection bias may arise because SEER covers approximately 28% of the U.S. population but is not a random sample; our analytic cohort was further restricted by inclusion/exclusion criteria (e.g., complete data on key variables), which could introduce non‑random selection and affect generalizability. Second, despite adjusting for a range of demographic, tumour, and facility‑level factors, residual confounding remains a concern. SEER does not capture potentially important confounders such as smoking history, performance status, detailed comorbidity indices, or specific biomarkers, all of which could influence treatment decisions and outcomes. Third, misclassification of treatment variables is possible because SEER records treatment as received within the first course of therapy, but lacks details on dosing, duration, regimen modifications (59). For example, systemic therapy is aggregated into broad categories, which may attenuate or distort true treatment effects. Fourth, as a retrospective registry‑based study, our analysis is inherently subject to data collection biases, missing data not at random, and the absence of patient‑reported outcomes, quality‑of‑life measures, and long‑term functional status. These limitations are intrinsic to the use of real‑world administrative data and should be considered when interpreting our results, particularly regarding causal inferences. Finally, dichotomisation may lead to loss of information, reduce statistical efficiency, and the chosen cut‑offs may not be generalisable to external populations. We therefore emphasise that our results should be interpreted as exploratory and hypothesis‑generating, and that the categorised age variable should be considered one of several stratification tools rather than a definitive biological threshold. Despite the above-mentioned limitations, our study has also identified multiple prognosis factors for PTCSLMs patients. Conclusions To our knowledge, this is the first study that systematically and extensively explore the independent prognostic factors of PTCSLMs. Our study suggested that the higher age at diagnosis and higher pathological grade were correlated with worse prognosis. In contrast, surgery, radiotherapy and chemotherapy showed a favorable prognostic association. In summary, age at diagnosis, histological type, T stage, grade stage, surgery, radiotherapy and chemotherapy are associated with prognosis and they are independent prognostic factors for PTCSLMs. Supplementary The article’s supplementary files as Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0333/prf