Human immunodeficiency virus (HIV)-associated primary effusion lymphoma (PEL) is a rare and aggressive subtype of B-cell non-Hodgkin lymphoma (NHL).1, 2 PEL usually arises in body cavities but may also present as extracavitary lesions without effusion (solid variant). Only a limited number of studies and case series on HIV–PEL have been published but factors associated with outcomes are insufficiently characterized, and standard treatment has not yet been established.3-8 In order to further define clinical characteristics, treatment and outcomes of HIV–PEL, we analyzed individual patient data from the prospective German HIV-related lymphoma cohort study. Adult HIV-1-infected patients with biopsy- or cytologically proven malignant lymphoma diagnosed at 34 participating centres in Germany and Austria from January 2005 to November 2018 were included in the study.9 Ethics approval was obtained from the ethics committees of the University of Cologne, and written informed consent was given by each participating patient. For statistical analyses, we used IBM SPSS Statistics software (version 27·0, IBM, Armonk, NY, USA) and GraphPad Prism (version 7.0d, San Diego, CA, USA). Patients’ causes of deaths were compared using a two-sided Fisher’s exact test. Differences between subgroups were assessed with the log-rank test. P values lower than 0·05 were considered statistically significant. Of 439 patients, 192 (44%) were diagnosed with diffuse large B-cell lymphoma (DLBCL), 136 (32%) with Burkitt lymphoma (BL), 53 (12%) with plasmablastic lymphoma (PBL), 13 (3%) with PEL and 45 (10%) had NHL not further classifiable or rare subtypes. All PEL patients were male, and the median age at PEL diagnosis was 51·0 years. The median time from first positive HIV test to PEL diagnosis was 5·7 years, and the median CD4+ T cell count was 195/µl (Table SI). Twelve of 13 patients (92%) had stage IV lymphoma and six patients (46%) were diagnosed with a solid variant (Table I). 1x DHAP 1x EPOCH 6x CHOP 1x mitoxantrone 4x R-Borte/ liposomal doxorubicin Liposomal doxorubicin 2x EPOCH → PD GMALL → CR 1x R-DHAP, 3x ICE, HD-BEAM/ASCT brentuximab, lenalidomide 1x CHOP 3x DA-EPOCH First-line chemotherapy consisted of cyclophosphamide, doxorubicin, vincristine, (etoposide), prednisolone [CHO(E)P]-like regimens (n = 9), liposomal doxorubicin for concurrent Kaposi sarcoma (n = 2), rituximab/bortezomib/liposomal doxorubicin (n = 1) and intrapleural mitoxantrone (n = 1; Table I). There were three complete responses (CR; 23%) and four partial responses (31%). Eight patients (62%) received salvage chemotherapy as outlined in Table I. However, a sustained remission was achieved in only one of the cases following six cycles of etoposide, prednisolone, vincristine, cyclophosphamide and doxorubicin (EPOCH). After a median follow-up of 12 months, two patients (15%) were alive and in first or second CR respectively. The median time to death was 12 months (range, 0–22) and the two survivors were alive at 6·1 and 10·7 years. Lymphoma in both surviving patients belonged to the classic variant with pleural effusions as the only sites of involvement. The two-year overall survival (OS) rate was 15·4% (95% confidence interval [95% CI] 0–35·0) as compared to 69·8% in DLBCL (95% CI 62·9–76·7), 73·1% in BL (95% CI 65·3–80·9), and 61·9% in PBL (95% CI 48·2–75·6, P < 0·0001; Fig 1A). Median OS of patients with PEL and PBL was 0·98 and 7·27 years, respectively, and it was not reached in DLBCL and BL. There was no statistically significant difference between patients with the classical and the solid PEL variant (median OS: 1·33 years and 0·92 years respectively; Fig 1B). The causes of death are listed in Table I. One of six patients (17%) with classic HIV–PEL died of refractory lymphoma compared to five of six patients (83%) with a solid variant (P = 0·08). One of the patients who died of infectious causes was in CR at the time of death. The prognosis of HIV–PEL remains poor in the combined antiretroviral therapy (cART) era. In the present study, only two patients were alive at last follow-up. Notably, all patients responding to first-line chemotherapy had pleural effusions with or without ascites as only sites of involvement, indicating a somewhat better prognosis of the classic PEL variant compared to solid variants. Another classic variant patient responded to B-ALL/B-NHL (GMALL)-based10 salvage chemotherapy and died of unknown cause while being in CR while two patients died of sepsis following a first cycle of chemotherapy. Thus, only one of six patients (17%) with classic HIV–PEL died of refractory lymphoma compared to five of six patients (83%) with a solid variant. A previous study also found higher CR rates in patients with the classic variant compared to extracavitary HIV–PEL (62% vs 41%) while there was no difference in OS.5 By contrast, patients with an extracavitary variant (n = 8) appeared to have slightly better survival than patients with a classic variant of HIV–PEL (n = 29) in another study with a median survival of 11 months vs 3 months.11 The median CD4+ T cell count of 195/µl at PEL diagnosis indicates that, unlike in earlier studies,3, 4 HIV–PEL does not primarily occur in severely immunosuppressed patients. A recent study on 20 patients with HIV–PEL found the Epstein–Barr virus (EBV) status of the tumour and elevated IL-6 levels to be prognostic.12 Median OS for EBV-encoded small RNA (EBER)-positive cases was not reached, with a three-year cancer-specific survival of 64%, whereas median OS in EBV− PEL was 10·4 months, with no patients alive at three years. In our study, all patients with EBV-positive PEL died of lymphoma and one of the two survivors was EBV-negative. However, as EBER was evaluated in only eight samples meaningful conclusions cannot be drawn. Optimal treatment of HIV–PEL remains uncertain. Promising results have been reported for the use of modified EPOCH which resulted in a 54% CR rate in 19 patients treated with EPOCH with/without rituximab, bevacizumab or high-dose methotrexate.12 Also, bortezomib added to chemotherapy resulted in ongoing remissions in a small case series6 while it proved unsuccessful in another report on three cases.13 It is, thus, difficult to draw firm conclusions from these observations. Notably, high-dose chemotherapy with autologous stem-cell support (HDCT/ASCT) was not successful as salvage treatment, which is in line with a previous report.14 The small sample size and the uncontrolled design are important limitations of the present study. However, given the rarity of the disease, large prospective studies will hardly be feasible, and our analysis provides some additional new data on HIV–PEL. In conclusion, the OS of patients with HIV–PEL remains poor, although long-term survival can be achieved in selected cases. International collaborative studies may help to define the optimal treatment approach. Christoph Wyen has received funding from the German Federal Ministry of Education and Research (BMBF; 01KI1017). MH, CH and PS conceived and designed the study; MH, MM, CW, JB, HT, J-CW, TW, CH and PS provided patients; MH, CW, CH, and PS collected and assembled the data; MH, CH and PS analyzed and interpreted the data; all authors wrote the manuscript, had access to the primary clinical data and gave final approval of the manuscript. TW received consultancy fees and travel grants from Gilead Science, MSD, and Janssen Pharmaceuticals. All other authors report no conflicts of interest. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.