Is phrenic nerve-sparing worthwhile for phrenic-involved thymic epithelial tumors? A retrospective cohort study Authors: Junkan Zhu, Yongqiang Ao, Zhenyang Lin, Jiahao Jiang, Jian Gao, Yuansheng Zheng, Xifei Jiang, Wei Wu, Junzhen Liu, Rongkui Luo, Jihong Dong, Shuai Wang, Jianyong Ding Published: 2026 ABSTRACT AbstractObjectivesTo determine whether phrenic nerve (PN) sparing offers a worthwhile trade-off between functional preservation and oncologic safety compared with resection in PN-involved thymic epithelial tumors (TETs), and to establish an objective tool for postoperative respiratory risk stratification.MethodsIn this retrospective cohort of patients with TETs with PN invasion, 122 propensity score–matched cases were analyzed from 2016 to 2024. PN preservation or resection was determined intraoperatively based on the extent of invasion. The diaphragmatic elevation (DE) ratio was calculated as [(preoperative diaphragm–apex distance − postoperative distance)/preoperative distance] × 100% according to X-rays, and cutoff values were defined by respiratory complication risk. Perioperative and oncologic outcomes were compared between the PN-sparing and PN resection groups.ResultsOverall perioperative morbidity and recovery outcomes were comparable between the PN-sparing and PN-resection groups; however, tumor progression was significantly more frequent on the PN-sparing group (39.3% vs 23.0%), resulting in worse progression-free survival (PFS) (P = .027), although overall survival (OS) remained comparable in the 2 groups. Multivariable analysis identified PN preservation as an independent predictor of progression (P = .049). Furthermore, our receiver operating characteristic–derived DE classification scheme (mild, ≤15%; moderate, 15-25%; severe, ≥25%) predicted severe postoperative respiratory complications (P < .001).ConclusionsPN resection improves PFS without increasing severe perioperative complications, although OS is comparable to that with PN-sparing.Key Words: thymic epithelial tumor, phrenic nerve, respiratory function Superior PFS in the PN-resection group compared with the PN-sparing group. Central Message Phrenic nerve resection enhances progression-free survival without added perioperative risk, despite no difference in overall survival. Perspective This study demonstrates that phrenic nerve sparing is independently associated with a significantly higher risk of tumor progression. We also validated an ROC-based diaphragmatic elevation classification for risk stratification. Consequently, complete resection should be prioritized to ensure oncologic safety over transient nerve preservation. Thymic epithelial tumors (TETs) are the most common primary mediastinal neoplasms, with phrenic nerve (PN) invasion reported in 10 to 30% of advanced cases (Masaoka-Koga stage III-IV).1, 2, 3 Surgeons frequently face the dilemma of whether to preserve or resect the PN, balancing the potential benefits of functional preservation against the oncologic advantage of more extensive tumor removal.4,5 Previous studies addressing PN management in TETs are sparse, with small sample sizes and inconsistent conclusions regarding short-term perioperative outcomes and long-term survival, underscoring the need for more rigorous research.6, 7, 8, 9 PN resection inevitably causes unilateral diaphragmatic paralysis and compromises pulmonary capacity, but it aims to secure negative margins and reduce recurrence risk. Conversely, while PN preservation avoids deliberate nerve transection and attempts to maintain baseline anatomy, it raises severe concerns for incomplete resection and a higher risk of tumor progression.4,5,7, 8, 9, 10, 11 Current evidence regarding these competing priorities remains highly conflicting, complicating clinical decision making and patient counseling. To address these gaps, we conducted a rigorously matched retrospective analysis of TETs with documented PN involvement to evaluate the clinical trade-offs between PN preservation and resection.1, 2, 3,7 Additionally, we developed a receiver operating characteristic (ROC)-derived diaphragmatic elevation (DE) classification system to provide an objective framework for guiding delayed therapeutic interventions. Rather than dictating a rigid surgical mandate, our study provides robust contemporary evidence to facilitate shared decision making, helping surgeons and patients carefully balance permanent functional trade-offs with local oncologic control. This cohort study has been reported in line with the STROCSS guidelines.12 Methods Study Design and Patients This retrospective single-center cohort study was conducted using the prospective mediastinal tumor database of Zhongshan Hospital, Fudan University and screened TET patients with PN invasion who underwent surgery at the institution between March 2016 and March 2024. Inclusion criteria were (1) pathologic diagnosis of TETs according to World Health Organization criteria (2021 edition); (2) pT2-4NxMx TETs according to the American Joint Committee on Cancer/Union for International Cancer Control/International Association for the Study of Lung Cancer/International Thymic Malignancy Interest Group TNM staging system, ninth edition13, 14, 15; (3) PN invasion confirmed by preoperative chest computed tomography (CT) and/or magnetic resonance imaging (MRI) and intraoperative findings; (4) treatment reviewed by a multidisciplinary mediastinal tumor board; and (5) medical records collected in accordance with the International Thymic Malignancy Interest Group’s standard definitions and policies.16 Exclusion criteria were (1) age >75 years; (2) Eastern Cooperative Oncology Group Performance Status >2; (3) history of other malignancies; (4) multiple surgeries for TETs; and (5) incomplete data. Patients with TETs with preoperative radiologic DE were excluded as well. The study was approved with a waiver of informed consent by the Institutional Review Board of Zhongshan Hospital, Fudan University on June 21, 2025 (No. B2025-333R) and was conducted in accordance with the Declaration of Helsinki. Surgery Detailed surgical protocols have been published previously.17,18 For minimally invasive procedures, we used a modified subxiphoid video-assisted thoracoscopic surgery approach. This technique involves a subxiphoid incision combined with a dedicated sternal retractor to elevate the sternum, providing excellent bilateral anterior mediastinal exposure while minimizing intercostal nerve trauma. Conversely, for large or locally advanced tumors, an open median sternotomy or clamshell incision was performed. This open approach is essential to ensure optimal vascular control, direct surgical visualization, and safe mobilization of complex mediastinal masses. Irrespective of the approach, extended thymectomy was performed according to the Masaoka criteria, involving complete removal of the thymus and surrounding anterior mediastinal adipose tissue.11 Locally invaded adjacent structures—including the lung parenchyma, pericardium, mediastinal pleura, chest wall, and great vessels (eg, superior vena cava)—were meticulously resected en bloc and reconstructed when necessary. Intraoperative management of PN was dictated strictly by the extent of tumor invasion. The decision to preserve or resect inherently relied on the primary surgeon's subjective visual assessment of invasion depth, tumor extent, and patient-specific factors (eg, myasthenia gravis [MG] status, baseline pulmonary reserve) (Figure 1). En bloc PN resection was performed when the tumor completely encased the nerve (>50% circumference) or when inseparable invasion of adjacent critical structures made preservation oncologically infeasible. Conversely, PN preservation was pursued under specific conditions. For minimal tumor involvement (PN adherent to the tumor edge without encasement), PN-sparing perineural skeletonization was performed to achieve macroscopic clearance.19 Furthermore, preservation was prioritized in scenarios where safety or physiologic stability outweighed local radicality. If absolute complete clearance was anatomically impossible (eg, extensive aortic encasement) and only macroscopic debulking was achievable, the PN was deliberately preserved to maintain respiratory mechanics. Similarly, in patients with MG, bilateral PN resection was strictly prohibited, and unilateral preservation was prioritized for those with generalized symptoms. Finally, during all nerve-sparing procedures, radiopaque titanium clips were routinely placed to precisely mark the preserved PN bed and other adjacent close margins to guide subsequent postoperative radiotherapy (Figure E1). Surgical procedures for phrenic nerve (PN) sparing and PN resection, with the PN highlighted by white frames. A, Invaded PN before preservation. B, Careful dissection of the PN. C, Complete preservation of intact PN. D, Invaded PN before resection. E, Intraoperative dissection and resection of the PN. F, Two nerve stumps after resection. Clip-guided postoperative radiotherapy following phrenic nerve (PN)-sparing surgery. A, Postoperative computed tomography scan displaying titanium clips strategically placed during the operation to mark the preserved PN bed and adjacent high-risk surgical margins. B, Radiotherapy planning and target volume delineation precisely guided by the intraoperatively placed titanium clips. Diaphragmatic Elevation The DE ratio was calculated as [(preoperative diaphragm–chest apex distance - postoperative distance)/preoperative distance] × 100% based on chest radiographs obtained during follow-up,8 which were derived from chest X-rays taken within 3 months postoperatively (Figure 2). The receiver operating characteristic (ROC) analyses were performed to identify optimal DE thresholds, using the incidence of respiratory complications as outcomes, including all-grade respiratory complications and grade III-IV respiratory complications. Thus, 2 cut-off values were derived from ROC analyses and subsequently applied for stratification of postoperative respiratory complication risk. Representative chest X-rays. A, Normal, left side (diaphragmatic elevation [DE] ratio = [D1-D2]/D1; 3.87%). B, Normal, right side (DE ratio, 1.93%). C, Mild DE, left side (DE ratio, 7.96%). D, Mild DE, right side (DE ratio, 10.1%). E, Moderate DE, left side (DE ratio, 22.7%). F, Moderate DE, right side (DE ratio, 16.6%). G, Severe DE, left side (DE ratio, 41.2%). H, Severe DE, right side (DE ratio, 50.3%). Postoperative Therapy and Follow-up All resected margins and structures were carefully marked and submitted for histopathologic examination. Patients with intraoperative hemodynamic instability were routinely admitted to the intensive care unit (ICU). The use of adjuvant therapy was individualized. Patients with incomplete resection routinely received postoperative chemotherapy and/or radiotherapy. For patients with complete resection, adjuvant radiotherapy was recommended for those with advanced-stage disease, high-grade thymoma, or thymic carcinoma, while adjuvant chemotherapy was considered for those with advanced-stage thymic carcinoma. Postoperative surveillance followed a standardized schedule, with clinic visits at 1 month and 3 months after surgery. For thymoma with PN invasion, chest CT was performed every 6 months during the first 2 years and annually thereafter. For thymic carcinoma with PN invasion, chest CT was performed every 3 to 6 months during the first 2 years and annually thereafter.13 At each follow-up visit, patients underwent clinical assessment and chest imaging, with neurologic evaluation for those with MG. Telephone follow-up was provided when necessary. MRI was considered in patients with a contraindication to contrast media or when minimizing radiation exposure was required. Positron emission tomography–CT scans were performed when necessary to provide a more comprehensive view of metabolic activity. MG was classified according to the Myasthenia Gravis Foundation of America’s clinical classification system. For MG patients, a standardized multidisciplinary protocol was implemented incorporating targeted preoperative immunomodulation (eg, intravenous immunoglobulin, plasmapheresis, efgartigimod) for high-risk cases, with restricted use of intraoperative nondepolarizing muscle relaxants. Rather than routine ICU admission, postoperative critical care and extubation were highly individualized and risk-stratified based on our recently established clinical classification for thymoma-associated MG.20 Prognosis also was assessed based on the postintervention status defined by this system, with overall remission defined as the combination of complete stable remission, pharmacologic remission (PR), and minimal manifestation status. Myasthenic crisis was defined as a rapid clinical decline in respiratory function necessitating mechanical ventilation (requiring prolonged postoperative intubation >24 hours) or reintubation after thymectomy.21 Progression-free survival (PFS) was evaluated from the date of surgery to the time of last follow-up or tumor progression, as diagnosed by biopsy or imaging during follow-up. Deaths without recurrence were considered censored observations. Overall survival (OS) was evaluated from the date of surgery to the time of death or last follow-up. Statistical Analysis Continuous variables were compared using the independent-samples t test, and categorical variables were analyzed with the Fisher exact test. To minimize baseline imbalances, propensity score matching (PSM) was performed at a 1:1 ratio using a greedy nearest-neighbor algorithm with a caliper width of 0.20, incorporating age, tumor histology, surgical approaches, TNM staging, preoperative radiotherapy/chemotherapy, and postoperative radiotherapy/chemotherapy. Survival outcomes were estimated by the Kaplan-Meier method and compared with the log-rank test. Risk factors for severe (grade III-IV) respiratory complications were identified using logistic regression. Prognostic variables for PFS were first evaluated in univariable Cox regression, and those with P < .05 were subsequently included in multivariable models. All statistical analyses were 2-sided, with significance defined as P < .05, and were conducted using R version 4.4.1 (R Foundation for Statistical Computing). Results Baseline Characteristics The screening flowchart is shown in Figure E2. A total of 221 patients with TETs and intraoperative PN invasion were included, of whom 108 (48.9%) underwent PN-sparing surgery and 113 (51.1%) underwent PN resection. After 1:1 PSM, 61 matched pairs were analyzed (Figure E3). Baseline characteristics were well balanced between the 2 groups (Table 1). Flowchart of patient screening. Distribution plot depicting the propensity score of unmatched and matched patients in the phrenic nerve (PN)-sparing and PN resection cohorts. Table 1Baseline data and perioperative outcomesCharacteristic/outcomeBefore propensity score matchingAfter propensity score matchingPN-sparing (N = 108)PN resection (N = 113)P valuePN-sparing (N = 61)PN resection (N = 61)P valueBaseline characteristics Sex, n (%).265>.999 Male55 (50.9)67 (59.3)34 (55.7)34 (55.7) Female53 (49.1)46 (40.7)27 (44.3)27 (44.3) Age, y, mean ± SD52.1 ± 12.651.7 ± 11.9.82451.9 ± 12.452.4 ± 12.7.801 CCI, n (%).613.810 017 (15.7)19 (16.8)9 (14.8)7 (11.5) 126 (24.1)19 (16.8)13 (21.3)10 (16.4) 229 (26.9)34 (30.1)18 (29.5)21 (34.4) ≥336 (33.3)41 (36.3)21 (34.4)23 (37.7) ECOG PS, n (%).231.278 084 (77.8)76 (67.3)047 (77.0)41 (67.2) 122 (20.4)33 (29.2)114 (23.0)18 (29.5) 22 (1.9)3 (2.7)20 (0.0)2 (3.3) 40 (0.0)1 (0.9) Surgery, n (%)<.001.807 Open thymectomy73 (67.6)104 (92.0)50 (82.0)52 (85.2) Thoracoscopic thymectomy35 (32.4)9 (8.0)11 (18.0)9 (14.8) Neoadjuvant radiotherapy, n (%).016.557 No100 (92.6)91 (80.5)53 (86.9)56 (91.8) Yes8 (7.4)22 (19.5)8 (13.1)5 (8.2) Neoadjuvant chemotherapy, n (%).009.698 No82 (75.9)66 (58.4)43 (70.5)40 (65.6) Yes26 (24.1)47 (41.6)18 (29.5)21 (34.4) Adjuvant radiotherapy, n (%)>.999>.999 No52 (48.1)55 (48.7)27 (44.3)28 (45.9) Yes56 (51.9)58 (51.3)34 (55.7)33 (54.1) Adjuvant chemotherapy, n (%).049>.999 No98 (90.7)91 (80.5)53 (86.9)54 (88.5) Yes10 (9.3)22 (19.5)8 (13.1)7 (11.5) MG, n (%)19 (17.6)17 (15.0).71612 (19.7)7 (11.5).318Tumor characteristics Histology, n (%)<.001.863 A6 (5.6)3 (2.7)4 (6.6)1 (1.6) AB22 (20.4)4 (3.5)5 (8.2)4 (6.6) B16 (5.6)8 (7.1)5 (8.2)7 (11.5) B1/B23 (2.8)1 (0.9)2 (3.3)1 (1.6) B224 (22.2)24 (21.2)14 (23.0)15 (24.6) B2/B317 (15.7)13 (11.5)7 (11.5)10 (16.4) B38 (7.4)21 (18.6)7 (11.5)5 (8.2) TC22 (20.4)39 (34.5)17 (27.9)18 (29.5) TNM stage, n (%)<.001.776 II70 (64.8)27 (23.9)27 (44.3)27 (44.3) IIIA11 (10.2)38 (33.6)10 (16.4)13 (21.3) IIIB1 (0.9)4 (3.5)1 (1.6)2 (3.3) IVA26 (24.1)44 (38.9)23 (37.7)19 (31.1) Lesion side, n (%).247.855 Right47 (43.5)59 (52.2)27 (44.3)25 (41.0) Left61 (56.5)54 (47.8)34 (55.7)36 (59.0)Structures co-resected, n (%)69 (63.9)106 (93.8)<.00148 (78.7)54 (88.5).221 Pulmonary42 (38.9)84 (74.3)<.00130 (49.2)37 (60.7).275 Diapraghm1 (0.9)4 (3.5).371 (1.6)4 (6.6).365 Pericardium55 (50.9)95 (84.1)<.00138 (62.3)47 (77.0).115 Vagus nerve0 (0.0)2 (1.8).4980 (0.0)1 (1.6)>.999 Superior vena cava3 (2.8)36 (31.9)<.0013 (4.9)9 (14.8).128 Innominate vein13 (12.0)48 (42.5)<.00112 (19.7)16 (26.2).518 Azygos1 (0.9)3 (2.7).6221 (1.6)0 (0.0)>.999PN, Phrenic nerve; SD, standard deviation; CCI, Charlson Comorbidity Index; ECOG PS, Eastern Cooperative Oncology Group Performance Status; MG, myasthenia gravis. Baseline data and perioperative outcomes PN, Phrenic nerve; SD, standard deviation; CCI, Charlson Comorbidity Index; ECOG PS, Eastern Cooperative Oncology Group Performance Status; MG, myasthenia gravis. Evaluation of Respiratory Function Short-term percentage declines (within 1 year) in forced vital capacity (12.3 ± 14.5% vs 18.6 ± 17.2%; P = .333) and forced expiratory volume in 1 second (13.8 ± 16.3% vs 20.1 ± 14.8%; P = .622) were numerically smaller in the PN-sparing group, but the differences lacked statistical significance. Long-term functional declines also remained comparable in the 2 groups (Figure 3, A). Respiratory function and oncologic outcomes in the phrenic nerve (PN)-sparing and PN-resection groups. A, Declining rates of forced vital capacity (FVC) and forced expiratory volume in 1 second (FEV1) in the PN-sparing (red) and PN resection (blue) groups over short-term and long-term follow-up. B, Longitudinal changes in diaphragmatic elevation (DE) in the 2 groups. C, Receiver operating characteristic curves of DE cutoff values for predicting respiratory complications (cutoff 1 for all-grade respiratory complications and cutoff 2 for grade III-IV respiratory complications). D, Distribution of DE severity categories (mild, moderate, severe) in the PN-sparing and PN resection groups. E, Progression-free survival in the 2 groups. F, Overall survival in the 2 groups. DE was quantified using a validated method described by Ventura and colleagues8, providing a standardized measure of diaphragmatic dysfunction. Longitudinal analysis of chest radiographs revealed distinct patterns: the PN-resection group exhibited progressive DE, peaking at 30% by 18 months postoperatively, whereas the PN-sparing group maintained stable and lower DE (5-15%) throughout follow-up (Figure 3, B). To translate these findings into clinically actionable metrics, we performed ROC analysis using grade I-IV and grade III-IV respiratory complications as endpoints, identifying optimal DE cut-offs at 16.17% (area under the curve, 0.729) and 27.42% (area under the curve, 0.702), respectively (Figure 3, C). For practical application, these thresholds were rounded to define 3 DE severity tiers: mild (≤15%), moderate (15-25%), and severe (≥25%). The incidence of respiratory complications increased progressively across DE classifications. Grade I-IV respiratory complications occurred in 57.4% of patients with mild DE, in 76.0% with moderate DE, and in 88.6% with severe DE. Similarly, the rate of grade III-IV respiratory complications rose with DE severity, from 2.5% of patients with mild DE to 24.0% with moderate DE and 38.6% with severe DE. Based on this classification model, we also discovered that PN-sparing was associated predominantly with mild DE (58.8% mild, 27.5% moderate, 13.7% severe), whereas PN resection was largely characterized by severe DE (17.2% mild, 19.0% moderate, 63.8% severe) (Figure 3, D). Perioperative Outcomes and Respiratory Complications After PSM, complete resection rates were comparable between the PN-sparing and PN resection groups (67.2% vs 72.1%; P = .215). Among the R2 resections, 93.8% (n = 30 of 32) were performed in patients with stage IVA disease. Regarding perioperative outcomes, there were no statistically significant differences in hospital length of stay or ICU length of stay. Furthermore, respiratory parameters—including the incidence of grade I-II and grade III-IV respiratory complications, as well as the requirement for and duration of postoperative mechanical ventilation—were comparable in the 2 group. Univariate logistic regression identified thymic carcinoma (P = .100) and DE stratification (P < .001) as potential predictors of severe respiratory complications. Subsequent multivariate analysis confirmed DE stratification as the sole independent predictor of grade III-IV respiratory complications (P < .001) (Table E1). Neurologic Outcomes Following PSM, the prevalence of preoperative MG, the preoperative medication for MG patients, and severity of MG were comparable in the 2 groups (19.7% in PN-sparing vs 11.5% in PN resection) (Tables E2 and E3). At the last follow-up, 58.3% (n = 7 of 12) of PN-sparing patients achieved remission (16.7% complete stable remission, 41.7% PR), compared with 28.6% (n = 2/7) of PN resection patients (14.3% PR, 14.3% minimal manifestations) (Figure E4). Among MG patients, the 2 groups did not differ significantly in any outcomes (Table E3). All instances of postoperative mechanical ventilation in MG patients were due to myasthenic crises. Improvement of myasthenia gravis in the peripheral nerve (PN)-sparing and PN resection cohorts, stratified by the Myasthenia Gravis Foundation of America classification. CSR, complete stable remission; PR, pharmacologic remission; MM, minimal manifestations; IM, improved; WS, worsening. Survival Analysis Tumor progression occurred in 24 patients (39.3%) in the PN-sparing group and 14 patients (23.0%) in the PN resection group. The pattern of progression was similar in the 2 groups, with pleural dissemination the most common type in both groups. Although the incidences of local progression, lymph node metastasis, and distant metastasis did not differ significantly, a trend toward a higher rate of pleural dissemination and distant metastasis was noted in the PN-sparing group (P = .089 and .057, respectively) (Table 2). Kaplan-Meier analysis demonstrated a significantly higher risk of tumor progression in the PN-sparing group (P = .027), while OS remained comparable between the groups (P = .504) (Figure 3, E and F). Subgroup analysis further revealed that PN resection significantly improved PFS in TNM stage II-III (P = .047) but showed no such benefit in stage IV (Figure E5). Both PN preservation and R2 resection were identified as potential predictors of tumor progression in univariate models, and PN preservation was confirmed as the sole independent predictor in multivariate analysis (P = .049) (Table E4). Table 2Perioperative patient outcomesOutcomePN-sparing (N = 61)PN resection (N = 61)P valueRadical surgery, n (%).215 R041 (67.2)44 (72.1) R11 (1.6)4 (6.6) R219 (31.1)13 (21.3)Hospital LOS, d, mean ± SD7.6 ± 6.010.2 ± 11.5.119ICU LOS, d, mean ± SD1.4 ± 2.43.3 ± 8.4.084Postoperative ventilation, n (%).078 No58 (95.1)51 (83.6) Yes3 (4.9)10 (16.4)Ventilation time, mean ± SD0.2 ± 0.81.3 ± 4.8.084Respiratory complications, n (%) I-II43 (70.5)51 (83.6).132 III-IV9 (14.8)17 (27.9).122Progression mode, n (%) Local progression7 (11.5)7 (11.5)>.999 Lymph node metastasis6 (9.8)5 (8.2)>.999 Pleural dissemination19 (31.1)10 (16.4).089 Distant metastasis5 (8.2)0 (0.0).057PN, Phrenic nerve; LOS, length of stay; SD, standard deviation; ICU, intensive care unit. Perioperative patient outcomes PN, Phrenic nerve; LOS, length of stay; SD, standard deviation; ICU, intensive care unit. Kaplan-Meier analysis of progression-free survival (PFS) stratified by TNM stage. A, Comparison of PFS between phrenic nerve (PN)-sparing and resection groups in patients with TNM stage II-III disease. B, Comparison of PFS between the PN-sparing and PN resection groups in patients with TNM stage IVA disease. Discussion Through PSM analysis of 122 TET cases with PN invasion, we found that PN resection provides significantly superior PFS without increasing short-term severe respiratory morbidity in our cohort. Given the genuine oncologic risk of residual disease associated with nerve preservation, these findings suggest that en bloc resection offers a meaningful local control advantage. In parallel, we developed the first ROC-based DE classification system tailored for TETs, which may provide an objective framework for guiding delayed therapeutic interventions. The PN provides the sole motor supply to the ipsilateral hemidiaphragm, making its integrity critical for normal respiratory function. PN transection reduces lung capacity and poses significant risks, particularly in patients with limited pulmonary reserve or MG.4,22, 23, 24 Previous studies have reported that PN preservation maintains significantly higher preoperative vital capacity compared to resection.22,24 In our cohort, short-term postoperative declines in forced vital capacity and forced expiratory volume in 1 second were numerically smaller in the PN-sparing group (12-14%) compared to the PN resection group (18-20%). While our data suggest that these differences tend to attenuate over time owing to physiologic compensation, this finding must be interpreted with strict caution. Crucially, our study lacks patient-reported outcomes and quality of life evaluations. Therefore, although objective pulmonary volumes partially recover, it remains entirely possible that patients undergoing PN resection continue to experience subjective dyspnea or diminished functional capacity, representing a notable limitation of this analysis. While prior studies link PN resection to increased respiratory morbidity,7,25 our propensity score–matched analysis demonstrated acceptable overall perioperative safety. Notably, surgical approach influenced this tolerance; minimally invasive techniques appeared to mitigate the physiologic impact of PN resection, whereas open sternotomy was associated with prolonged hospital stay. Furthermore, despite historical concerns, PN resection in the high-risk MG subgroup did not significantly increase acute myasthenic crises, although PN-sparing yielded higher long-term MG remission rates.26,27 Ultimately, while open surgery necessitates careful perioperative management, these short-term physiologic considerations should not deter surgeons from prioritizing en bloc PN resection for maximal oncologic clearance in well-selected candidates. Clinically, PN function can be evaluated using intraoperative compound muscle action potential monitoring, nerve conduction studies, ultrasound assessment of diaphragmatic motion, or imaging-based measurements such as the diaphragmatic height index.28, 29, 30 Among these, postoperative assessment of DE on chest X-ray is the most convenient and widely accessible method. Unilateral PN resection causes ipsilateral diaphragmatic paralysis and DE, which can markedly impair respiratory mechanics and contribute to postoperative complications.31,32 Because traditional fixed cut-offs often fail to account for individual anatomic variation, we used ROC-derived thresholds (15% and 25%) to objectively quantify DE.33,34 This approach demonstrated that the extent of DE is directly correlated with the risk of severe respiratory complications. Furthermore, our data show that DE after PN resection is progressive—peaking at approximately 30% by 18 months—underscoring the importance of extended respiratory follow-up. To ensure unbiased quantification for this analysis, patients who underwent concurrent diaphragmatic plication were strictly excluded. By defining the DE thresholds associated with severe respiratory morbidity, our findings may provide a theoretical reference for the postoperative evaluation of delayed therapeutic plication; however, future multicenter prospective studies are needed to validate the real-world clinical utility of this classification system in guiding surgical interventions.35 While PN preservation avoids permanent paralysis, its oncologic safety remains heavily debated.4,10,19 Our analysis definitively clarifies this risk; multivariable Cox regression identified PN preservation as an independent predictor of tumor progression, resulting in significantly worse PFS. Attempting to peel a tumor off the nerve inherently compromises the local surgical margin; however, we must acknowledge that isolated microscopic margins rarely capture the full clinical picture in our cohort. PN invasion frequently coexists with aggressive stage IVA disease. In these scenarios, achieving an absolute R0 clearance is often anatomically impossible, necessitating planned macroscopic cytoreduction at other disseminated sites. Because final pathologic staging reflects the worst overall margin, localized microscopic disease left at the preserved nerve bed is frequently masked by gross residual disease elsewhere. Nevertheless, our multivariable data confirm that regardless of this complex margin distribution, leaving an involved nerve in situ fundamentally compromises oncologic radicality. Therefore, when PN involvement is unequivocal, en bloc resection provides a definitive local control benefit.9,11,36 While PN resection improved PFS, it lacked a demonstrable OS benefit. More importantly, the acceptable perioperative tolerance observed in our cohort lacks broad generalizability. For patients with marginal baseline function, the toll of unilateral paralysis poses a potentially prohibitive risk. Recognizing the true burden of these trade-offs requires prioritizing patient-reported outcomes and quality of life in broader populations. Ultimately, the decision for en bloc resection requires shared decision making, balancing local oncologic control against individualized cardiopulmonary reserves rather than applying a universal surgical mandate. Several limitations of this study merit consideration. First, our retrospective design introduces confounding by indication. The intraoperative decision to preserve the PN relied on the surgeon's real-time assessment of tumor invasion and patient physiology. This highly individualized triage means that the PN-sparing group likely represents less aggressive local disease. Second, lacking standardized pulmonary function testing precluded precise functional quantification, a common challenge in retrospective studies.23,37 Third, excluding patients age >75 years or with Eastern Cooperative Oncology Group Performance Status >2 limits generalizability. Although necessary to prevent baseline frailty from confounding our respiratory data, this selection bias means that our recommendation for complete resection cannot be blindly extrapolated to vulnerable populations, who strictly require case-by-case evaluation. Finally, our single-center retrospective setting highlights the need for multi-institutional prospective trials. Incorporating patient-reported outcomes, quality-of-life assessments, and advances in intraoperative nerve monitoring will help clarify whether the objective functional benefits of PN preservation truly translate into durable, subjective well-being.38,39 In conclusion, in patients with adequate cardiopulmonary reserve, en bloc PN resection improves PFS without increasing severe perioperative respiratory morbidity. However, because this approach does not demonstrate an OS benefit and carries lifelong DE, the surgical strategy necessitates meticulous patient counseling and shared decision making. Additionally, our ROC-derived DE classification may provide an objective, evidence-based tool to help guide delayed therapeutic plication for persistent and severe DE; however, future multicenter prospective studies are required to validate its real-world clinical utility. Webcast You can watch a Webcast of this AATS meeting presentation by going to: https://www.aats.org/resources/is-phrenic-nerve-sparing-worth-12445. Conflict of Interest Statement The authors reported no conflicts of interest. The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest. Footnotes Drs Zhu, Ao, and Lin contributed equally to this work. This study was supported by the National Natural Science Foundation of China (Grants 82404059 and 82472924). The study protocol was approved by the Ethics Committee of Zhongshan Hospital, Fudan University on June 21, 2025 (approval B2025-333R) and was conducted under the guidance of the Declaration of Helsinki. Contributor Information Shuai Wang, Email: wang.shuai@zs-hospital.sh.cn. Jianyong Ding, Email: ding.jianyong@zs-hospital.sh.cn. Appendix E1 Table E1Univariable and multivariable logistic regression analyses evaluating grade III-IV respiratory complicationsVariableUnivariable analysisMultivariable analysisOR (95% CI)P valueOR (95% CI)P valueAge1.00 (0.96-1.03).822--Sex (female vs male)0.48 (0.18-1.19).123--CCI (≥3 vs <3)1.14 (0.45-2.76).774--Myasthenia gravis (yes vs no)1.92 (0.61-5.51).240--Histologic type (TC vs thymoma)0.38 (0.10-1.10).1000.44 (0.11-1.41).193Lesion side (right vs left)0.83 (0.35-2.02).682--Surgery (thoracoscopic thymectomy vs open surgery)0.91 (0.24-2.78).876Resection (R2 vs R0/R1)0.61 (0.19-1.67).364--DE stratification3.63 (1.90-7.98)< .0013.50 (1.84-7.67)< .001OR, Odds ratio; CI, confidence interval; CCI, Charlson Comorbidity Index; TC, thymic carcinoma; DE, diaphragmatic elevation. Univariable and multivariable logistic regression analyses evaluating grade III-IV respiratory complications OR, Odds ratio; CI, confidence interval; CCI, Charlson Comorbidity Index; TC, thymic carcinoma; DE, diaphragmatic elevation. Table E2Preoperative medications for MG patientsAdministrationsPN sparing (N = 19)PN resection (N = 17)P valuePyridostigmine bromide, n (%).255 ≤60 mg thrice daily16 (84.2)11 (64.7) ≥120 mg thrice daily3 (15.8)6 (35.3)Oral prednisone acetate, n (%).589 <30 mg once daily13 (68.4)6 (35.3) ≥30 mg once daily2 (10.5)2 (11.8)Intravenous glucocorticoid pulse, n (%)1 (5.3)2 (11.8).593Tacrolimus, n (%).560 ≤2 mg/d8 (42.1)3 (17.6) >2 mg/d2 (10.5)2 (11.8)Mycophenolate mofetil, n (%)1 (5.3)1 (5.9)>.999Azathioprine, n (%)3 (15.8)0 (0.0).231Other immunosuppressants, n (%)4 (21.1)1 (5.9).342Targeted biologics (eg, efgartigimod), n (%)3 (15.8)0 (0.0).231Intravenous immunoglobulin, n (%)13 (68.4)10 (58.8).730Plasmapheresis, n (%)2 (10.5)1 (5.9)>.999 Preoperative medications for MG patients Table E3Perioperative outcomes of patients with MGOutcomePN-sparing (N = 12)PN resection (N = 7)P valueOsserman grade, n (%).417 I3 (25.0)2 (28.6) IIa7 (58.3)2 (28.6) IIb2 (16.7)3 (42.9)MGFA grade, n (%).534 I3 (25.0)2 (28.6) IIa4 (33.3)1 (14.3) IIb3 (25.0)1 (14.3) IIIa2 (16.7)1 (14.3) IIIb0 (0.0)2 (28.6)Myasthenic crisis, n (%)4 (33.3)4 (57.1).377Hospital LOS, d, mean ± SD6.9 ± 2.816.9 ± 23.1.300ICU LOS, d, mean ± SD2.6 ± 2.211.7 ± 16.9.203Postoperative ventilation, n (%).377 No8 (66.7)3 (42.9) Yes4 (33.3)4 (57.1)Ventilation time, mean ± SD0.3 ± 0.97.6 ± 12.9.188Respiratory complications, n (%) I-II7 (58.3)7 (100).106 III-IV3 (25.0)3 (42.9).617MG, Myasthenia gravis; PN, phrenic nerve; MGFA, Myasthenia Gravis Foundation of America; LOS, length of stay; SD, standard deviation; ICU, intensive care unit. Perioperative outcomes of patients with MG MG, Myasthenia gravis; PN, phrenic nerve; MGFA, Myasthenia Gravis Foundation of America; LOS, length of stay; SD, standard deviation; ICU, intensive care unit. Table E4Univariable and multivariable Cox proportional hazards analyses evaluating PFSVariableUnivariable analysisMultivariable analysisHR (95% CI)P valueHR (95% CI)P valueAge1.01 (0.98-1.04).520--Sex (female vs male)1.02 (0.54-1.96).943--CCI (≥3 vs <3)1.00 (0.51-1.96).999--MG (yes vs no)0.68 (0.26-1.74).419--Histologic type (TC vs thymoma)1.46 (0.76-2.82).260--PN (resection vs sparing)0.48 (0.25-0.93).0310.51 (0.26-1.00).049Lesion side (left vs right)0.59 (0.31-1.12).105--Resection (R2 vs R0/R1)2.00 (1.03-3.90).0421.84 (0.94-3.61).074Adjuvant radiotherapy (yes vs no)0.71 (0.37-1.36).304--Adjuvant chemotherapy (yes vs no)1.33 (0.55-3.19).526--PFS, Progression-free survival; HR, hazard ratio; CI, confidence interval; CCI, Charlson Comorbidity Index; MG, myasthenia gravis; TC, thymic carcinoma; PN, phrenic nerve. Univariable and multivariable Cox proportional hazards analyses evaluating PFS PFS, Progression-free survival; HR, hazard ratio; CI, confidence interval; CCI, Charlson Comorbidity Index; MG, myasthenia gravis; TC, thymic carcinoma; PN, phrenic nerve. References