Clinical outcomes of non-traumatic chylothorax at a Singapore tertiary referral hospital Authors: Brian Lee Wei Chua, Si Ling Young, Qiao Li Tan, Carrie Kah-Lai Leong, Jane Jing Yi Wong, Ivana Gilcrist Chiew Sian Phua, Wen Ting Lim, Ken Junyang Goh Published: 2026 ABSTRACT AbstractBackground and objectiveWhile the management of traumatic chylothorax has advanced over the past two decades, outcomes for non-traumatic chylothorax remain poor, with a paucity of data to guide management and prognostication.MethodsWe performed a retrospective analysis of patients diagnosed with chylothorax between January 2020 and December 2024. Clinical data including interventions and clinical outcomes were recorded.ResultsWe evaluated 40 patients with a median age of 61 (IQR: 50–72) years. Cancer was the most common cause (67.5%), of which 18 (45%) patients had solid-organ malignancies and 9 (22.5%) had haematological malignancies. Other aetiologies included drug-induced (10%) and liver cirrhosis (7.5%). Despite dietary measures implemented in 65.0% of patients, 52.5% of patients required ≥2 pleural interventions, with chest drain and indwelling pleural catheter insertions performed in 32 (80.0%) and 8 (20.0%) patients, respectively. Lymphatic imaging was infrequently (20.0%) performed, with chyle leak identified in only one patient. The 90-day and 1-year mortality was 37.5% and 55.0%, respectively, with the highest 1-year mortality rates seen with solid organ cancers (83.3%) and liver cirrhosis (66.7%). Eighteen (45.0%) patients had clinical resolution of chylothorax, and this was highest in patients with drug-induced chylothorax (100.0%) followed by patients with haematological malignancies (66.7%).ConclusionNon-traumatic chylothorax secondary to haematological malignancies and drugs have more favourable outcomes compared with solid organ cancers and liver cirrhosis. This highlights the need for a personalised and multidisciplinary approach to the management of chylothorax. WHAT IS ALREADY KNOWN ON THIS TOPIC Outcomes for non-traumatic chylothorax remain poor, with a paucity of data to guide management and prognostication. Outcomes for non-traumatic chylothorax remain poor, with a paucity of data to guide management and prognostication. WHAT THIS STUDY ADDS Our study sheds more light on the clinical characteristics, treatment course and clinical outcomes of patients with non-traumatic chylothorax based on their underlying aetiology. Our study sheds more light on the clinical characteristics, treatment course and clinical outcomes of patients with non-traumatic chylothorax based on their underlying aetiology. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Our study suggests that a personalised approach is required when managing non-traumatic chylothorax. More data are required in well-designed prospective studies to improve patient outcomes. Our study suggests that a personalised approach is required when managing non-traumatic chylothorax. More data are required in well-designed prospective studies to improve patient outcomes. Introduction Chylothorax refers to the accumulation of chyle within the pleural cavity.1 Chyle is a milky fluid formed in the small intestines during digestion and contains triglyceride, fat-soluble vitamins, lymphocytes and immunoglobulins.1 Chylothorax is rarely seen, accounting for approximately 3% of all pleural effusions.2 The causes of chylothorax can be broadly classified into traumatic and non-traumatic aetiologies, with most cases being traumatic.3 While outcomes for traumatic chylothorax have improved with interventions like thoracic duct ligation or embolisation, the same cannot be said for non-traumatic chylothorax, where reported 1-year mortality rates remain approximately 50%.46 Malignancy, in particular, haematological cancer like lymphoma, is the most common cause of a non-traumatic chylothorax.4 Poor prognostic markers include malignancies, bilateral effusions and chronicity of chylous effusions.7 A consensus statement was recently published to address pertinent questions regarding management of chylothorax.8 However, much of the evidence to guide treatment of chylothorax is based on traumatic chylothorax. In contrast, recommendations for the management of non-traumatic chylothorax are based largely on small case series. Expert centres have explored the use of MRI-guided intervention algorithms in the management of non-traumatic chylothorax, but this has not been widely adopted as such capabilities may not be widely available.9 Furthermore, management should not be based on a ‘one-size-fits-all’ approach as the pathophysiology of non-traumatic chylothorax can be driven by different mechanisms.2 10 Unfortunately, there is a paucity of data, especially in the Asian population.6 11 Evaluating clinical characteristics of non-traumatic chylothorax and understanding differences in outcomes and prognosis will be valuable in enabling personalised treatment decisions in non-traumatic chylothorax. Therefore, we sought to evaluate the clinical characteristics and outcomes of patients with non-traumatic chylothorax in an Asian tertiary referral hospital. We present this manuscript in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting checklist. Methods We performed a retrospective analysis of all consecutive patients who were diagnosed with chylothorax between January 2020 and December 2024. The study was conducted at Singapore General Hospital, the largest tertiary referral hospital in Singapore with a 2000-bed capacity. Clinical data including patient demographics, aetiologies, pleural fluid characteristics, diagnostic imaging, interventions and clinical outcomes were recorded. The diagnosis of chylothorax was established based on pleural fluid triglyceride levels >1.24 mmol/L and pleural fluid cholesterol levels <5.18 mmol/L on initial and subsequent pleural fluid analysis.12 The categorisation of pleural fluid as exudative or transudative was based on Light’s criteria. The number of hospitalisation days were calculated based on hospital admissions that were directly related to patient’s underlying chylothorax. Clinical resolution was defined as survival to clinical and radiological stability with no further pleural interventions required. Management of non-traumatic chylothorax was decided by a multidisciplinary team involving primary physician, pleural unit and interventional radiologist (IR). Decisions were mainly guided by symptoms and drain output. IR was consulted regarding feasibility and modality of lymphatics imaging if drain output remains high despite optimal medical therapy. As this study was performed with anonymised data retrieved from a clinical audit, it was exempted from formal institutional review board approval. Statistics Continuous variables are expressed as median and IQR and categorical variables as numbers with percentages. All statistical analyses were performed using SPSS Statistics software V.22.0 (IBM Corp, Armonk). Patient and public information There was no patient or public involvement in the design, conduct and reporting of this work. Results Patient characteristics Forty patients were included during the study period, and their characteristics are summarised in table 1. The median age for our study population was 61 (IQR: 50–72) years. Twenty-seven (67.5%) patients were female. Cancer was the most common cause (27 patients or 67.5%) of non-traumatic chylothorax, of which 18 (45%) patients had solid-organ malignancies and 7 (22.5%) patients had haematological malignancies. The most common solid-organ malignancy was breast cancer with seven (17.5%) patients, followed by lung and pancreatic cancer with three (7.5%) patients each. The rest of the patients had intra-abdominal malignancies including hepatocellular carcinoma, rectal, ovarian, cervical and gastric cancers. Among the nine patients with haematological malignancies, all of them had lymphoma. In our cohort, four (10.0%) patients had drug-induced chylothorax, of which three were attributed to Dasatinib and one was related to a trial drug for cancer treatment. Of the six (15.0%) patients with other causes of chylothorax, two were diagnosed with lymphangioleiomyomatosis (LAM), one had Gorham-Stout disease, and the remaining three were attributed to fluid overload. Table 1Baseline characteristics of study populationN=40Age, years61 (50–72)Gender, female27 (67.5)Causes of chylothorax Solid organ malignancies*18 (45.0) Haematological malignancies*9 (22.5) Liver cirrhosis3 (7.5) Drug-induced4 (10.0) Others6 (15.0)Site of chylothorax Right24 (60.0) Left10 (25.0) Bilateral6 (15.0)Appearance of pleural fluid Milky37 (92.5) Serous3 (7.5)Exudative pleural effusion29 (72.5)Pleural fluid LDH (mmol/L)320 (202–538)Pleural fluid protein (g/L)40 (35–46.9)Pleural fluid triglycerides (mmol/L)2.9 (1.7–5.1)Pleural fluid cholesterol (mmol/L)2 (0.9–2.6)Ultrasound findings Echogenic appearance10 (25.0) Pleural/diaphragmatic nodules2 (5.0)Data are presented in number (percentage) and median (IQR) for categorical and continuous variables, respectively.*Breast (seven), lung (three), pancreas (three), liver (one), rectal (one), ovarian (one), cervical (one), gastric (one).LDH, lactate dehydrogenase. Baseline characteristics of study population Data are presented in number (percentage) and median (IQR) for categorical and continuous variables, respectively. Breast (seven), lung (three), pancreas (three), liver (one), rectal (one), ovarian (one), cervical (one), gastric (one). LDH, lactate dehydrogenase. Pleural fluid characteristics The pleural fluid characteristics are summarised in tables1 2. Most patients had right-sided (60.0%) chylothorax and six (15.0%) patients had bilateral chylothoraxes. Of the six patients with bilateral effusions, four patients had haematological malignancies. Milky pleural fluid was seen in 37 (92.5%) patients and most pleural effusions were exudative (72.5%) based on Light’s criteria. Diagnostic ultrasound findings were mostly non-specific, with 10 (25.0%) patients having an echogenic appearance and 2 (5.0%) patients with diaphragmatic or pleural nodules seen on ultrasound. Table 2Clinical characteristics of chylothorax within each categorySolid organ malignancies (N=18)Haematological malignancies (N=9)Liver cirrhosis(N=3)Drug induced(N=4)Others(N=6)Age61 (50–79)58 (47–70)72 (64–79)63 (60–70)54 (33–75)Site Right14 (77.8)2 (22.2)2 (66.7)2 (50.0)4 (66.7) Bilateral04 (44.4)01 (25.0)1 (16.7)Exudative pleural effusion13 (72.2)7 (77.8)1 (33.3)3 (75.0)5 (83.3)Pleural fluid LDH (mmol/L)315 (196–677)518 (320–538)226 (206–477)269 (263–378)248 (113–2140)Pleural fluid protein (g/L)43 (37–45)33.5 (33–41.5)10 (10–20)49 (39–49)43 (40–47)Pleural fluid triglycerides (mmol/L)2.4 (1.6–3.2)3.45 (2.3–7.3)3.3 (2.8–3.9)4.1 (2.7–5.6)5.6 (1.6–16.6)Pleural fluid cholesterol2.2 (0.9–2.6)1.8 (0.9–2.1)0.5 (0.5–0.5)1.6 (1.4–2.0)2.8 (1.7–3)Data are presented in number (percentage) and median (IQR) for categorical and continuous variables, respectively.LDH, lactate dehydrogenase. Clinical characteristics of chylothorax within each category Data are presented in number (percentage) and median (IQR) for categorical and continuous variables, respectively. LDH, lactate dehydrogenase. Clinical interventions The clinical interventions performed for our study population are summarised in table 3. Despite being one of the cornerstones of the management of chylothorax, only 26 (65.0%) patients complied with prescribed low-fat diet with medium-chain triglyceride supplementation. Five (12.5%) patients were given a trial of pharmacological therapy. Physicians opted for either intravenous or subcutaneous octreotide, with doses ranging from 50 to 100 μg every 8 hours, and they were all terminated after a week due to persistently high drain output. Of these five patients, three patients eventually had clinical resolution after definitive treatment. Table 3Clinical interventions within each categoryAll (N=40)Solid organ malignancies (N=18)Haematological malignancies (N=9)Liver cirrhosis (N=3)Drug induced (N=4)Others (N=6)Dietary measures26 (65.0)13 (72.2)7 (77.8)2 (66.7)04 (66.7)Pharmacological therapy5 (12.5)1 (5.6)3 (33.3)001 (16.7)Number of pleural interventions2 (1–2)2 (1–2)2 (1–3)2 (1–2)2 (1–2)2 (1–2)Thoracocentesis13 (32.5)6 (33.3)1 (11.1)1 (33.3)3 (75.0)2 (33.3)Pleural drain insertion32 (80.0)15 (83.3)9 (100.0)2 (66.7)15 (83.3)Numbers of days drain kept in-situ22 (8–29)22 (11–24)28 (6–46)19 (14–34)149 (16–114)Indwelling pleural catheter8 (20.0)3 (16.7)3 (33.3)1 (33.3)01* (16.7)Number of days IPC kept in-situ70 (20–75)70 (45–100)75 (43–178)29075Medical thoracoscopy5 (12.5)2 (11.1)1 (11.1)02 (50.0)0Talc pleurodesis13 (32.5)7 (38.9)3 (33.3)00 (0)3 (50.0)Number of patients who underwent lymphatic imaging8 (20.0)3 (16.7)3 (33.3)01 (25.0)1 (16.7) Lymphoscintigraphy3 (7.5)2 (11.1)0001 (16.7) CT/MR lymphangiography3 (7.5)02 (22.2)01 (25.)0 Conventional lymphangiogram5 (12.5)2 (11.1)2 (22.2)001 (16.7) Leak identified1 (2.5)01 (11.1)000Surgical intervention5 (12.5)1 (5.6)01 (33.3)03 (50.0)Data are presented in number (percentage) and median (IQR) for categorical and continuous variables, respectively.*Case of suspected lymphoma but declined diagnostic evaluation in view of poor functional status.IPC, indwelling pleural catheter. Clinical interventions within each category Data are presented in number (percentage) and median (IQR) for categorical and continuous variables, respectively. Case of suspected lymphoma but declined diagnostic evaluation in view of poor functional status. IPC, indwelling pleural catheter. The median number of pleural interventions performed per patient was 2 (IQR: 1–2) procedures. The most common procedure was pleural drain insertion with 32 (80.0%) patients, followed by thoracocentesis with 13 (32.5%) patients. The median number of days the pleural drains were left in situ was 22 (IQR: 8–29) days. Eight (20.0%) patients underwent indwelling pleural catheter (IPC) insertion with a median duration of 70 (IQR: 20–75) days. None of the patients achieved spontaneous autopleurodesis. Thirteen (30.0%) patients underwent talc pleurodesis, of which 10 cases were for malignant effusions. The remaining three patients were diagnosed with LAM and Gorham-Stout disease and underwent talc pleurodesis as part of their surgery. Eight (20.0%) patients had lymphatic imaging performed, of which three patients underwent two different modalities of imaging. Conventional lymphangiogram was performed in five (12.5%) with chyle leak identified in one (2.5%) patient. Five (12.5%) patients underwent surgery as part of their management of chylothorax with thoracic duct ligation performed in one patient. One patient with metastatic pancreatic cancer with concomitant malignant chylous ascites underwent surgical repair of diaphragmatic defect. Another patient with Child’s C Liver Cirrhosis underwent liver transplant and surgical repair of underlying diaphragmatic defect. The remaining two patients with LAM had chylopneumothorax and underwent video-assisted thoracoscopic surgery (VATS) bullectomy and talc pleurodesis. All five patients had clinical resolution on subsequent follow-ups. Clinical outcomes Clinical outcomes are summarised in table 4. The median length of hospitalisation was 31 (IQR: 14–60) days. Of note, patients with drug-induced chylothorax only required a median hospitalisation period of 4 (IQR: 3–7) days. In our study cohort, there were 15 (37.5%) deaths at 90 days and 22 (55.0%) deaths at 1-year mark. 90-day and 1-year mortality rates were the highest in patients with solid organ cancer and liver cirrhosis. Of the 15 (83.3%) patients with solid organ malignancies who passed away at the 1-year mark, five patients were diagnosed with breast cancer, three had lung cancer, two had gynaecological cancer, two had pancreatic cancer. The remaining three patients had rectal cancer, hepatocellular carcinoma, gastric cancer. Eighteen (45.0%) patients had clinical resolution, and this was highest in patients with drug-induced chylothorax (100.0%) followed by patients with haematological malignancies (66.7%). Table 4Clinical outcomes within each categoryAll(N=40)Solid organ malignancies(N=18)Haematological malignancies(N=9)Liver cirrhosis (N=3)Drug induced(N=4)Others(N=6)P valueNumber of hospitalisation days31 (14–60)30 (18–54)44 (31–68)45 (41–135)4 (3–7)70 (15–114)Clinical resolution18 (45.0)5 (27.8)6 (66.7)1 (33.3)4 (100.0)3 (50.0)0.06090-day mortality15 (37.5)12 (66.7)2 (22.2)1 (33.3)000.0091-year mortality22 (55.0)15 (83.3)3 (33.3)2 (66.7)02 (33.3)0.008 Clinical outcomes within each category Discussion In this study, we examined the clinical characteristics, aetiologies and clinical outcomes of patients with non-traumatic chylothorax. To our knowledge, this is the first observational study reporting outcomes of patients with non-traumatic chylothorax from a Asian centre.6 12 Malignancy was the most common cause of chylothorax in our study, with lymphoma being the most common malignancy. This is similar to other studies where lymphoma accounted for up to 50% of non-traumatic chylothorax cases.1315 However, solid organ cancer as a group (45.0%) was more common than lymphoma (22.5%) in our study population, with metastatic breast cancer the most common. There is little data published regarding the relationship between different solid-organ malignancies and non-traumatic chylothorax. Most studies report postsurgical complications following oesophagectomy and lung cancer resections.16 17 Interestingly, among the 11 (27.5%) patients with a transudative effusion, 7 patients were found to have underlying malignant pleural effusions. In a study by Porcel et al, which included 67 patients with non-traumatic chylothorax, three out of six patients with transudative effusions were also found to have underlying cancer.6 These highlight the importance of evaluating for cancer in patients with chylothorax, even with transudative effusions. Most of our patients (92.5%) had milky pleural fluid. While other studies have reported less than 50% of milky pleural fluid appearance, their studies comprise largely of traumatic chylothorax with pre and post-operative dietary restrictions.14 18 19 In the study by Porcel et al, this percentage (73%) is closer to our study population.6 However, it is also plausible that testing for pleural fluid triglycerides was not performed in patients with non-milky pleural effusions from chylothorax due to the lack of clinical suspicion. In terms of interventions and management, there was significant heterogeneity among our patients. Only 65% of our patients adopted dietary measures. In non-traumatic chylothorax, the role of nutritional support remains unclear as the data have been limited to small case series.20 In a recently published statement, dietary measures were recommended for traumatic chylothorax, but only suggested as a consideration in patients with non-traumatic chylothorax.10 More data are required to establish the role and effectiveness of dietary measures in non-traumatic chylothorax.21 22 This is especially pertinent as many patients who have underlying cancer are already at high risk of malnutrition.23 It remains unclear how best to personalise treatment of chyle leak diets with regards to duration and follow-up. Furthermore, difficulty with adherence to the strict dietary requirements for a chyle leak diet can be difficult to sustain for many patients, leading to poor compliance rates, such as those seen in our study.6 Majority of our patients required more than one pleural procedure and up to 80% of our patients required a chest drain during their hospitalisation, with a median duration of chest drainage of 22 days. Eight patients eventually had IPCs inserted, with a median duration of 70 days. These reflect the poorer prognosis of non-traumatic chylothorax, with clinical resolution seen in only less than half (45.0%) of our patients—very similar to other studies of predominantly non-traumatic chylothorax cases.6 24 This is likely to reflect how treatable the underlying cause is, with very low rates of clinical resolution seen with solid organ cancers (27.8%) and liver cirrhosis (33.3%) in our study population, compared with a 100% resolution rate in drug-induced chylothorax and 66.7% in patients with lymphoma. Despite recommendations for dynamic lymphatics imaging in patients with refractory chylothorax,8 10 the number of patients who eventually undergo lymphangiogram is low. In the study by Porcel et al, only 3% of patients had lymphangiograms performed and none had thoracic duct embolisation or ligation.6 Similarly, even with resources to perform lymphatic imaging in our institution, only 20% of patients underwent lymphatic imaging. The reason for this is unclear, but plausibly related to a more conservative approach for patients with metastatic solid organ cancer, adopting a watch and wait approach while monitoring their response to treatment.10 20 There are also lower success rates of thoracic duct embolisation for non-traumatic chylothorax compared with traumatic aetiologies.2527 The role of routine lymphatic imaging for all patients is unclear. In a recent study by Goh et al involving 20 patients with non-traumatic chylothorax, no significant abnormalities to direct further lymphatic intervention were found in 11 out of 12 patients who underwent lymphatic imaging with MR lymphangiogram.28 Nevertheless, of the five patients who underwent conventional lymphangiography in our institution, four achieved clinical resolution despite only one patient having an identifiable leak. Lymphangiogram alone has been shown to be effective in up to 30%–70% of cases.26 27 More studies are required to evaluate the role of lymphatic imaging and intervention in non-traumatic chylothorax. In our study population, 90-day and 1-year mortality rates were 37.5% and 55%, respectively, and the median length of hospital stay was 31 (IQR: 14–60) days. This highlights the significant morbidity and mortality in patients with chylothorax and are similar to other published studies on non-traumatic chylothorax.6 24 While factors like malignancy have been proposed to be poor prognostic markers for clinical resolution and survival, there appears to be significant differences in these outcomes depending on the specific cancer type.7 29 In our study, patients with solid organ cancers appear to have significantly worse rates of clinical non-resolution and mortality compared with patients with lymphoma. Again, these highlight that conditions which are more amenable to treatment, such as lymphoma, drug-induced chylothorax and LAM, are more likely to experience clinical improvement and resolution of the chylothorax. The same observations are seen with rates of clinical resolution of chylothorax. Management of chylothorax should, therefore, take on a personalised approach, taking into consideration the underlying aetiology (eg, the type of cancer), volume of chyle leak and other pertinent clinical factors. Limitations There are limitations to our study. It is a single-centre study with a relatively small population size, and our results may not be generalisable to other countries or institutions. Although management was a multidisciplinary effort, decisions made regarding treatment can be variable between different physicians. Nevertheless, our study provides useful data about the characteristics of non-traumatic chylothorax based on different aetiology. To our knowledge, this is the first study that examines adult patients with non-traumatic chylothorax only, providing much-needed data regarding this very rare pleural disease. Conclusion In conclusion, our study demonstrated the heterogeneity of clinical characteristics, management and outcomes of non-traumatic chylothorax. Non-traumatic chylothorax secondary to haematological malignancies, drugs and congenital lymphatic anomalies appear to have favourable clinical resolution and mortality rates, compared with solid organ cancers and liver cirrhosis. A personalised and multidisciplinary approach is required when managing chylothorax. More data are required in well-designed prospective studies to improve the management of non-traumatic chylothorax.