Article Text

Original research
Brain metastasis and survival outcomes after first-line therapy in metastatic melanoma: a multicenter DeCOG study on 1704 patients from the prospective skin cancer registry ADOREG
  1. Cindy Franklin1,2,
  2. Peter Mohr3,
  3. Leonie Bluhm3,
  4. Friedegund Meier4,
  5. Marlene Garzarolli4,
  6. Michael Weichenthal5,
  7. Katharina Kähler5,
  8. Imke Grimmelmann6,
  9. Ralf Gutzmer7,
  10. Jochen Utikal8,
  11. Patrick Terheyden9,
  12. Rudolf Herbst10,
  13. Sebastian Haferkamp11,
  14. Claudia Pfoehler12,
  15. Andrea Forschner13,
  16. Ulrike Leiter13,
  17. Fabian Ziller14,
  18. Frank Meiss15,
  19. Jens Ulrich16,
  20. Alexander Kreuter17,
  21. Christoffer Gebhardt18,
  22. Julia Welzel19,
  23. Bastian Schilling20,
  24. Martin Kaatz21,
  25. Anca Sindrilaru22,
  26. Edgar Dippel23,
  27. Dorothee Nashan24,
  28. Michael Sachse25,
  29. Carsten Weishaupt26,
  30. Harald Löffler27,
  31. Thilo Gambichler28,
  32. Carmen Loquai29,30,
  33. Lucie Heinzerling31,
  34. Stephan Grabbe30,
  35. Dirk Debus32,
  36. Gaston Schley33,
  37. Jessica C Hassel34,
  38. Gerhard Weyandt35,
  39. Maike Trommer36,
  40. Georg Lodde37,
  41. Jan-Malte Placke37,
  42. Lisa Zimmer37,
  43. Elisabeth Livingstone37,
  44. Jürgen Christian Becker37,38,
  45. Susanne Horn37,39,
  46. Dirk Schadendorf37 and
  47. Selma Ugurel37
  1. 1Department of Dermatology and Venereology, University of Cologne, Faculty of Medicine and University Hospital Cologne, Cologne, Germany
  2. 2Center for Integrated Oncology Aachen-Bonn-Cologne-Düsseldorf (CIO ABCD), Cologne, Germany
  3. 3Department of Dermatology, Elbe-Kliniken Buxtehude, Buxtehude, Germany
  4. 4Department of Dermatology, University Hospital Carl Gustav Carus, TU Dresden and, Skin Cancer Center at the University Cancer Center Dresden and National Center for Tumor Diseases (NCT), Dresden, Germany
  5. 5Department of Dermatology, Skin Cancer Center, Schleswig-Holstein University Hospital, Campus Kiel, Kiel, Germany
  6. 6Skin Cancer Center Hannover, Department of Dermatology, Hannover Medical School, Hanover, Germany
  7. 7Department of Dermatology, Muehlenkreiskliniken Minden and Ruhr University Bochum, Minden, Germany
  8. 8Skin Cancer Unit, German Cancer Research Center (DKFZ) and Department of Dermatology, Venereology and Allergology, University Medical Center Mannheim, University of Heidelberg, Mannheim, Germany
  9. 9Department of Dermatology, University of Lübeck and Schleswig-Holstein University Hospital, Campus Lübeck, Lübeck, Germany
  10. 10Department of Dermatology, HELIOS Klinikum Erfurt, Erfurt, Germany
  11. 11Department of Dermatology, University Hospital Regensburg, Regensburg, Germany
  12. 12Department of Dermatology, Saarland University Medical School, Homburg, Homburg/Saar, Germany
  13. 13Department of Dermatology, University Hospital Tübingen, Tübingen, Germany
  14. 14Department of Dermatology, DRK Hospital Chemnitz-Rabenstein, Chemnitz, Germany
  15. 15Department of Dermatology and Venereology, Medical Center, University of Freiburg, Freiburg im Breisgau, Germany
  16. 16Department of Dermatology and Skin Cancer Center, Harzklinikum Dorothea Christiane Erxleben, Quedlinburg, Germany
  17. 17Department of Dermatology, Venereology and Allergology, Helios St. Elisabeth Klinik Oberhausen, University Witten-Herdecke, Oberhausen, Germany
  18. 18Department of Dermatology and Venereology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany
  19. 19Department of Dermatology and Allergology, University Hospital Augsburg, Augsburg, Germany
  20. 20Department of Dermatology and Venereology, University Hospital Würzburg, Würzburg, Germany
  21. 21Department of Dermatology, SRH Wald-Klinikum Gera, Gera, Germany
  22. 22Department of Dermatology and Venereology, University Hospital Ulm, Ulm, Germany
  23. 23Department of Dermatology, Ludwigshafen Medical Center, Ludwigshafen, Germany
  24. 24Department of Dermatology, Hospital of Dortmund, Dortmund, Germany
  25. 25Skin Cancer Center, Department of Dermatology, Klinikum Bremerhaven Reinkenheide, Bremerhaven, Germany
  26. 26Department of Dermatology, University Hospital of Muenster, Muenster, Germany
  27. 27Department of Dermatology, SLK-Kliniken Heilbronn, Heilbronn, Germany
  28. 28Department of Dermatology, Ruhr-University Bochum, Bochum, Germany
  29. 29Department of Dermatology, Klinikum Bremen-Ost, Gesundheit Nord gGmbH, Bremen, Germany
  30. 30Department of Dermatology, University Medical Center of the Johannes Gutenberg University Mainz, Mainz, Germany
  31. 31Department of Dermatology and Allergology, Ludwig-Maximilian University, Munich, Germany
  32. 32Department of Dermatology, Nuremberg General Hospital, Paracelsus Medical University, Nuremberg, Germany
  33. 33Department of Dermatology and Venereology, Helios Klinikum Schwerin, Schwerin, Germany
  34. 34National Center for Tumor Diseases (NCT), Department of Dermatology, University Hospital Heidelberg, Heidelberg, Germany
  35. 35Department of Dermatology and Allergology, Hospital Bayreuth, Bayreuth, Germany
  36. 36Department of Radiation Oncology and Cyberknife Center, University of Cologne, Faculty of Medicine and University Hospital Cologne, Cologne, Germany
  37. 37Department of Dermatology, Venereology and Allergology, University Hospital Essen and German Cancer Consortium (DKTK) Partner Site Essen, Essen, Germany
  38. 38Translational Skin Cancer Research, German Cancer Consortium (DKTK), Deutsches Krebsforschungszentrum, Heidelberg, Germany
  39. 39Rudolf-Schönheimer-Institute of Biochemistry, Medical Faculty of the University Leipzig, Leipzig, Germany
  1. Correspondence to Dr Cindy Franklin; cindy.franklin{at}uk-koeln.de

Abstract

Background Despite the availability of effective systemic therapies, a significant number of advanced melanoma patients develops brain metastases. This study investigated differences in incidence and time to diagnosis of brain metastasis and survival outcomes dependent on the type of first-line therapy.

Methods Patients with metastatic, non-resectable melanoma (AJCCv8 stage IIIC–V) without brain metastasis at start of first-line therapy (1L-therapy) were identified from the prospective multicenter real-world skin cancer registry ADOREG. Study endpoints were incidence of brain metastasis, brain metastasis-free survival (BMFS), progression-free survival (PFS), and overall survival (OS).

Results Of 1704 patients, 916 were BRAF wild-type (BRAFwt) and 788 were BRAF V600 mutant (BRAFmut). Median follow-up time after start of 1L-therapy was 40.4 months. BRAFwt patients received 1L-therapy with immune checkpoint inhibitors (ICI) against CTLA-4+PD-1 (n=281) or PD-1 (n=544). In BRAFmut patients, 1L-therapy was ICI in 415 patients (CTLA-4+PD-1, n=108; PD-1, n=264), and BRAF+MEK targeted therapy (TT) in 373 patients. After 24 months, 1L-therapy with BRAF+MEK resulted in a higher incidence of brain metastasis compared with PD-1±CTLA-4 (BRAF+MEK, 30.3%; CTLA-4+PD-1, 22.2%; PD-1, 14.0%). In multivariate analysis, BRAFmut patients developed brain metastases earlier on 1L-therapy with BRAF+MEK than with PD-1±CTLA-4 (CTLA-4+PD-1: HR 0.560, 95% CI 0.332 to 0.945, p=0.030; PD-1: HR 0.575, 95% CI 0.372 to 0.888, p=0.013). Type of 1L-therapy, tumor stage, and age were independent prognostic factors for BMFS in BRAFmut patients. In BRAFwt patients, tumor stage was independently associated with longer BMFS; ECOG Performance status (ECOG-PS), lactate dehydrogenase (LDH), and tumor stage with OS. CTLA-4+PD-1 did not result in better BMFS, PFS, or OS than PD-1 in BRAFwt patients. For BRAFmut patients, multivariate Cox regression revealed ECOG-PS, type of 1L-therapy, tumor stage, and LDH as independent prognostic factors for PFS and OS. 1L-therapy with CTLA-4+PD-1 led to longer OS than PD-1 (HR 1.97, 95% CI 1.122 to 3.455, p=0.018) or BRAF+MEK (HR 2.41, 95% CI 1.432 to 4.054, p=0.001), without PD-1 being superior to BRAF+MEK.

Conclusions In BRAFmut patients 1L-therapy with PD-1±CTLA-4 ICI resulted in a delayed and less frequent development of brain metastasis compared with BRAF+MEK TT. 1L-therapy with CTLA-4+PD-1 showed superior OS compared with PD-1 and BRAF+MEK. In BRAFwt patients, no differences in brain metastasis and survival outcomes were detected for CTLA-4+PD-1 compared with PD-1.

  • melanoma

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

WHAT IS ALREADY KNOWN ON THIS TOPIC

  • Although systemic therapy of metastatic melanoma has advanced dramatically, it is unclear at present, which treatment strategy is preventing the development of brain metastases best and if that treatment is also associated with the best overall survival.

WHAT THIS STUDY ADDS

  • To address this, we assessed a prospectively collected real-world multicenter patient cohort with 1704 melanoma patients for the time until occurrence of brain metastases, progression-free and overall survival.

  • In our cohort, first-line therapy with BRAF+MEK targeted therapy led to faster development of brain metastases than PD-1±CTLA-4 immune checkpoint inhibition (ICI) in BRAF V600 mutant patients.

  • In BRAF V600 mutant patients, CTLA-4+PD-1 ICI resulted in better overall survival compared with PD-1 ICI or BRAF+MEK inhibitor therapy, while in BRAF wild-type patients there was no difference in the occurrence of brain metastases and overall survival between treatment with PD-1 monotherapy and CTLA-4+PD-1 combined ICI.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

  • Our findings underline that survival outcomes have to be assessed separately for BRAFmut and BRAFwt patients, and that ICI, particularly CTLA-4+PD-1, should be preferably chosen as first-line therapy in BRAFmut melanoma patients without MBM.

Background

Outcome of melanoma patients has improved dramatically in the last decade. Immune checkpoint inhibitors (ICIs) directed against CTLA-4 (ipilimumab) or PD-1 (nivolumab, pembrolizumab) or their combination can be applied as systemic therapies in all advanced melanoma patients. Melanoma patients with an activating BRAF V600 mutation (BRAFmut) can alternatively receive a targeted therapy (TT) with inhibitors of the RAS-RAF-MEK-ERK (MAPK) signaling pathway. The current standard of care is the combination of BRAF plus MEK inhibitors (BRAF+MEK), based on randomized phase III trials demonstrating improved survival of the combination compared with BRAF monotherapy.1–3 Monotherapy with ipilimumab (CTLA-4) has shown inferior response and survival rates compared with PD-1 monotherapy (PD-1) or the combined therapy with ipilimumab and nivolumab (CTLA-4+PD-1) and is therefore no longer considered standard of care.4 Recently, in a large retrospective study, our group could not detect differences in survival outcomes in 450 melanoma patients with brain metastases (MBM) who received different first-line therapies in addition to different types of radiotherapy.5 Prospective studies without additional locoregional therapies in MBM patients, as well as a meta-analysis, showed higher intracranial effectiveness and better survival outcomes with first-line therapy with CTLA-4+PD-1 compared with PD-1.6–8 In melanoma patients without MBM it is currently unclear, which treatment strategy is preventing the development of MBM best. This question is of utmost importance, since MBM account for around 50% of melanoma-related deaths.9 Recent results of two prospective studies have shown better overall survival (OS) with the use of ICIs as first-line and TT as second-line therapy in metastatic BRAFmut patients.10 11 At present, we do not know if TT or ICI as first-line therapy results in better prevention and delay of MBM and if this results in better survival. While several ongoing prospective randomized clinical trials address the question of the optimal sequence in metastatic patients, it has to be considered that a number of patients die—frequently by occurrence of MBM—reducing the chance to receive an effective second-line therapy.

The aim of this study was to compare survival outcomes and onset of MBM occurrence depending on different systemic treatment options (PD-1 or CTLA-4+PD-1 ICI, BRAF+MEK TT) in BRAF wild-type (BRAFwt) and BRAFmut melanoma patients starting their first-line therapy without MBM in a prospectively collected multicenter real-world patient cohort.

Patients and methods

Study design

Patients with advanced, non-resectable melanoma (AJCCv8 stage IIIC–IV) without brain metastasis at start of their first-line therapy (1L-therapy) treated with inhibitors of PD-1 (nivolumab, pembrolizumab), CTLA-4 (ipilimumab), the combination of both (ipilimumab+nivolumab), or BRAF+MEK (dabrafenib+trametinib, vemurafenib+cobimetinib, encorafenib+binimetinib) between January 2011 and January 2022 were identified from the prospective multicenter skin cancer registry ADOREG of the German Dermatologic Cooperative Oncology Group. Patients with ocular melanoma, patients with non-resectable stage IIIA/B disease, and patients who received BRAF inhibitors without a MEK inhibitor were excluded from this analysis. Data on patient and tumor characteristics, as well as baseline parameters of the first and second non-adjuvant systemic treatment were collected. Best overall response as assessed by the investigators was categorized as complete response, partial response, stable disease, mixed response and progressive disease according to RECIST V.1.1.12–14 Tumor stagings were performed at baseline and every 3 months thereafter by CT scan of the chest and abdomen and an MRI scan of the brain. Study endpoints were time until the first diagnosis of brain metastasis (brain metastasis-free survival, BMFS), progression-free (PFS) and OS.

Statistical analysis

Univariate and multivariate Cox proportional hazards regression analyses were performed to assess the impact of baseline patient and tumor characteristics and therapeutic measures on BMFS, PFS and OS. The following parameters were included into the univariate and multivariate analyses: sex, age, type of therapy, Eastern Cooperative Oncology Group performance status (ECOG-PS), lactate-dehydrogenase (LDH) serum levels, disease stage by AJCCv8 (IIIB/C, IV M1a, IV M1b, IV M1c), and previous adjuvant therapy with ICI or BRAF+MEK inhibitors. Median follow-up time was calculated as time from start of first non-adjuvant systemic therapy till death or last patient contact. BMFS was defined as time from start of first non-adjuvant systemic therapy until first detection of brain metastasis or last patient contact (censored BMFS), PFS as time from start of systemic therapy until disease progression or last patient contact (censored PFS) and OS as time from start of systemic therapy until death or last patient contact (censored OS). Kaplan-Meier estimates were used for BMFS, PFS and OS calculation; differences between groups were assessed by two-sided log-rank tests. P values <0.05 were considered statistically significant. Patients with missing data were excluded from the respective analyses. Statistical analyses were performed with IBM SPSS Statistics V.27.

Results

Patient characteristics

Data freeze was February 1, 2022. From 2243 patients identified in the ADOREG registry, 1871 met all study inclusion criteria. Additional 167 patients were excluded because of an unknown BRAF mutational status. A detailed study flow is provided in figure 1.

Figure 1

Study flow. A total of 2243 patients from 33 skin cancer centers were identified in the prospective multicenter ADOREG registry. Of these, 1704 patients could be included in the outcome analysis.

The resulting 1704 patients were used for all further analyses, and had a median follow-up time of 40.4 (range: 0.1–119.8) months after start of 1L-therapy. A total of 930 (54.6%) of these patients had received a subsequent second-line therapy. With regard to pretreatment, 67 (3.9%) patients had received an adjuvant therapy for stage III, and 127 (7.5%) patients for stage IV melanoma (three patients received both). For detailed patient characteristics see table 1.

Table 1

Baseline characteristics and therapy outcome of first-line therapy (all patients; n=1704)

A total of 788 patients were BRAFmut, and 916 were BRAFwt. As expected, BRAFwt patients were older and encomprised 64 (6.8%) patients with mucosal melanoma, compared with only 4 (0.5%) patients with mucosal melanoma in the BRAFmut patient group. 60.5% (n=477) of BRAFmut patients and 49.5% (n=453) of BRAFwt patients received a second-line therapy. Median OS was similar for BRAFmut and BRAFwt patients (36.44 vs 37.39 months, p=0.922).

BRAF wild-type patients

Two hundred and eighty-one (30.7%) BRAFwt patients received CTLA-4+PD-1, 544 (59.4%) PD-1, and 91 (9.7%) CTLA-4 ICI as 1L-therapy. For detailed characteristics of BRAFwt patients, see online supplemental table S1A. Patients who received CTLA-4+PD-1 were significantly younger than patients who received PD-1 (age ≤65 years: 56.2% vs 26.7%). Patients who received CTLA-4+PD-1 showed a better PS compared with those who received PD-1 monotherapy (ECOG-PS>0: 55.5% vs 48.0%), while patients with very high levels of serum LDH also received more often CTLA-4+PD-1 (serum LDH ≥2 x ULN: 31.0% vs 21.9%). In addition, patients who received combination ICI had more advanced disease stages (56.6% stage IV M1c) compared with patients who received PD-1 monotherapy (41.9% stage IV M1c). The objective response rate was 32.0% for CTLA-4+PD-1, 27.2% for PD-1, and 13.2% for CTLA-4. First-line therapy was stopped significantly more often because of toxicity in patients who received CTLA-4+PD-1 (32.0% vs 12.3%), whereas in patients who received PD-1 therapy, therapy was ended more often due to disease progression (27.4% vs 44.1%). Disease progression occurred in 47.0% (CTLA-4+PD-1), 60.3% (PD-1) and 75.8% (CTLA-4). At 24 months after start of 1L-therapy, the incidence of MBM was 14.2% with CTLA-4+PD-1, 16.0% with PD-1 and 24.2% with CTLA-4 (figure 2). Median OS was not reached for CTLA-4+PD-1, 34.56 (95% CI 28.00 to 41.12) months for PD-1, and 37.39 (95% CI 21.99 to 52.79) months for CTLA-4 (online supplemental table S1B).

Supplemental material

Figure 2

Development of brain metastasis on different types of first-line therapy in BRAF mutant and BRAF wild-type melanoma patients. (A, C) Incidence of brain metastasis at 24 months after start of first-line therapy, and (B, D) brain metastasis-free survival. The log-rank test was used to compare between groups; p<0.05 was considered significant.

Univariate Cox regression analysis showed an association of ECOG-PS, serum LDH, tumor stage, and type of 1L-therapy with OS, and of tumor stage and therapy type with PFS; for details see online supplemental table S2 and figure 3. For BMFS, only tumor stage was significantly prognostic (online supplemental table S3).

Figure 3

Kaplan-Meier curves showing progression-free and overall survival for first-line therapy in BRAF mutant and BRAF wild-type melanoma patients. (A, B) Progression-free survival; (C, D) overall survival. The log-rank test was used to compare between groups; p<0.05 was considered significant.

Multivariate Cox regression analysis in BRAFwt patients including age, gender, ECOG-PS, serum LDH, previous adjuvant therapy with ICI or BRAF+MEK, tumor stage, and type of 1L-therapy (CTLA+PD-1, PD-1 or CTLA-4; table 2) identified ECOG-PS>0 (0 vs 1: HR 1.821, 95% CI 1.350 to 2.456, p<0.001; 0 vs 2: HR 2.510, 95% CI 1.433 to 4.247, p=0.001), elevated serum LDH (not elevated vs elevated: HR 1.433, 95% CI 1.10 to 1.851, p=0.006), and tumor stage (IV M1c vs IV M1a: HR 0.558, 95% CI 0.320 to 0.972, p=0.039; IV M1c vs IV M1b: HR 0.625, 95% CI 0.454 to 0.862, p=0.002), but not the type of 1L-therapy (CTLA-4+PD-1 or PD-1) as independent prognostic factors for OS. First-line therapy with CTLA-4 (CTLA-4+PD-1 vs CTLA-4: HR 2.346, 95% CI 1.469 to 3.747, p<0.001, CTLA-4 vs PD-1: HR 0.466, 95% CI 0.299 to 0.727, p=0.001), and higher tumor stage (IV M1c vs IV M1b: HR 0.640, 95% CI 0.489 to 0.837, p=0.001) were independently prognostic for a shorter PFS. Only tumor stage could be identified as an independent prognostic factor for BMFS (stage IV M1c vs stage IIIC/D: HR 0.359, 95% CI 0.181 to 0.711, p=0.003; stage IV M1c vs stage IV M1a: HR 0.158, 95% CI 0.039 to 0.646, p=0.010; table 3).

Table 2

Multivariate Cox regression analysis for progression-free and overall survival in BRAF-wild-type patients

Table 3

Multivariate Cox regression analysis for brain metastasis-free survival in BRAF-wild-type patients

BRAF-mutant patients

Thirteen point seven % (n=108) of BRAFmut patients received CTLA-4+PD-1, 33.5% (n=264) received PD-1, 43 (5.5%) received CTLA-4, and 373 (47.3%) received BRAF+MEK as 1L-therapy (online supplemental table S4A). Of 373 patients who received BRAF+MEK as first-line therapy, 257 (68.9%) received dabrafenib+trametinib, 62 (16.6%) vemurafenib+cobimetinib and 54 (14.5%) encorafenib+binimetinib. Patients who received CTLA-4+PD-1 were significantly younger than patients with other therapy types (age ≤65 years: 77.8% CTLA-4+PD-1, 43.6% PD-1, 56.6% BRAF+MEK, and 60.5% CTLA-4), while patients treated with BRAF+MEK had significantly higher serum LDH levels (>2 × ULN: 40.8% BRAF+MEK, 29.6% CTLA-4+PD-1, 23.5% PD-1, and 32.6% CTLA-4). Patients who received CTLA-4+PD-1 or BRAF+MEK had more advanced disease stages than patients who received PD-1 or CTLA-4 monotherapy (stage IV M1c: 60.2% CTLA-4+PD-1, 57.4% BRAF+MEK, 36.4% PD-1, and 37.2% CTLA-4). Objective response rate was 36.4% for BRAF+MEK, 32.6% for PD-1, 28.7% for CTLA-4+PD-1, and 4.7% for CTLA-4, with a high number of best responses not reported. 26.9% of patients who received CTLA-4+PD-1 stopped therapy because of toxicity, compared with only 8.3% with PD-1, 17.4% with BRAF+MEK, and 14% with CTLA-4. In contrast, 31.5% of patients treated with CTLA-4+PD-1, 48.5% of patients treated with PD-1, 45.8% of patients treated with BRAF+MEK, and 37.2% of patients treated with CTLA-4 discontinued 1L-therapy because of disease progression. Disease progression occurred in 43.5% of patients who received CTLA-4+PD-1, in 58.7% of patients who received PD-1, in 56.0% of patients who received BRAF+MEK, and in 74.4% of patients who received CTLA-4. Median PFS was highest for BRAF+MEK TT (9.46; 95% CI 8.341 to 10.583 months) compared with CTLA-4+PD-1 (8.74; 95% CI 2.83 to 14.65) months), PD-1 (5.95 months; 95% CI 4.24 to 7.66), and CTLA-4 (2.14; 95% CI 2.048 to 2.223 months) ICI. Median OS was not reached for CTLA-4+PD-1 and for PD-1 monotherapy, and was 26.28 (95% CI 21.1 to 31.5) months for BRAF+MEK and 34.92 (95% CI 3.3 to 41.4) months for CTLA-4. At 24 months after start of 1L-therapy, BRAF+MEK TT revealed a higher incidence of MBM compared with PD-1±CTLA-4 ICI (CTLA-4+PD-1: 22.2%, PD-1: 14.0%, BRAF+MEK: 30.3%, CTLA-4: 44.2%, figure 2, online supplemental table S4B).

Univariate Cox regression analysis revealed ECOG-PS, serum LDH, tumor stage, and type of 1L-therapy as associated with PFS and OS; for details see online supplemental table S5 and figure 3. Age, tumor stage, and type of systemic therapy were associated with BMFS (online supplemental table S6).

In the multivariate Cox regression analysis (table 4), we detected ECOG-PS>0 (ECOG-PS=1: HR 1.740, 95% CI 1.221 to 2.479, p=0.001; ECOG-PS=2: HR 2.513, 95% CI 1.541 to 4.099, p<0.001), elevated serum LDH (HR 1.464, 95% CI 1.087 to 1.970, p=0.012), tumor stage of IV M1c (III C/D: HR 0.481, 95% CI 0.261 to 0.884, p=0.018; IV M1a: HR 0.450, 95% CI 0.262 to 0.774, p=0.004) as independently negatively associated with OS, and 1L-therapy with CTLA-4+PD-1 as independently positively associated with OS (PD-1: HR 1.969, 95% CI 1.122 to 3.455, p=0.018; CTLA-4: HR 3.948, 95% CI 1.743 to 8.942, p=0.001; BRAF+MEK: HR 2.409, 95% CI 1.432 to 4.054, p=0.001). ECOG-PS>0, elevated serum LDH, tumor stage of IV M1c, and 1L-therapy with CTLA-4 showed an independent negative association with PFS. Independent prognostic factors for longer BMFS found by multivariate analysis were age >65 years (HR=0.632 compared with age ≤65 years; 95% CI 0.431 to 0.927, p=0.019) and tumor stage IV M1a (HR 0.294, 95% CI 0.1133 to 0.651, p=0.003) compared with stage IV M1c, while 1L-therapy with BRAF+MEK was negatively associated with BMFS compared with CTLA-4+PD-1 (HR 0.60, 95% CI 0.332 to 0.945; p=0.030) and PD-1 (HR 0.575, 95% CI 0.372 to 0.888; p=0.013; table 5). A subgroup analysis revealed no differences for PFS or OS when comparing the three different BRAF+MEK combinations in the multivariate analysis (online supplemental table S11A,B).

Table 4

Multivariate Cox regression analysis for progression-free and overall survival in BRAF-mutant patients

Table 5

Multivariate Cox regression analysis for brain metastasis free-survival in BRAF-mutant patients

Brain metastases

BRAFmut patients developed MBM significantly more often than BRAFwt patients (27.4% (n=216) vs 18.7% (n=171) MBM, table 1). Median time till development of MBM was not reached in both groups, while mean time till development of MBM was 66.5 months for BRAFmut and 83.0 months for BRAFwt patients (p<0.001). With regard to 1L-therapy, 16.0% (n=45) of BRAFwt patients treated with CTLA-4+PD-1, 18.9% (n=103) of patients treated with PD-1, and 25.3% (n=23) of patients treated with CTLA-4 developed MBM during follow-up (online supplemental table S1A), compared with 25.0% (n=27) of BRAFmut patients treated with CTLA-4+PD-1, 17.0% (n=45) treated with PD-1, 33% (n=123) treated with BRAF+MEK and 48.8% (n=21) treated with CTLA-4 (online supplemental table S4A). In BRAFmut patients, median time till development of MBM was 51.8 months for BRAF+MEK, 39.2 months for CTLA-4 and not reached for CTLA-4+PD-1 or PD-1 (p<0.001), whereas it was not reached for BRAFwt patients on treatment with CTLA-4+PD-1, PD-1 or CTLA-4. In the Kaplan-Meier survival analysis, we noticed that in BRAFmut patients ICI therapy with CTLA-4+PD-1 resulted in a more rapid development of MBM compared with PD-1-treated patients, while this was not visible in the BRAFwt cohort (figure 2). No differences for BMFS where detected when comparing the three different BRAF+MEK combinations in the multivariate analysis.

Therapy sequence

Next we wanted to assess, if a certain therapy sequence of first-line and second-line therapy is associated with improved BMFS or OS. Altogether, 930 patients with known BRAF mutational status had received a second-line therapy. These patients were significantly younger than patients who did not receive second-line therapy. They also had more often received CTLA-4 or BRAF+MEK as 1L-therapy. Patients with second-line therapy developed MBM more often than patients without second-line therapy (for details, see online supplemental table S7). A total of 477 (51.3%) patients who received second-line therapy were BRAFmut, and 453 (48.7%) were BRAFwt (online supplemental table S8).

Multivariate Cox regression analysis including therapy sequence revealed ECOG-PS>0 (ECOG-PS=1: HR 1.745, 95% CI 1.090 to 2.795, p=0.021, ECOG-PS=2: HR 3.146, 95% CI 1.656 to 5.975, p<0.001) as negatively associated with OS for BRAFmut patients (online supplemental table S10). Additionally, the sequence of BRAF+MEK followed by CTLA-4+PD-1 was negatively associated with OS when compared with CTLA-4+PD-1 followed by BRAF+MEK (HR 1.988, 95% CI 1.026 to 3.852, p=0.042). For BMFS the sequence of BRAF+MEK followed by CTLA-4+PD-1 was inferior to PD-1 followed by BRAF+MEK. In BRAFwt patients, the sequence of CTLA+PD-1 and PD-1 did not affect OS or BMFS (online supplemental table S9).

Discussion

In our present study, we analyzed a large prospectively collected real-world cohort of advanced melanoma patients, who started non-adjuvant 1L-therapy in the absence of brain metastases. This specific setting was chosen to test for differences in time from therapy start until first diagnosis of brain metastases in different types of 1L-therapy. Our findings demonstrate that in BRAFmut patients 1L-therapy with BRAF+MEK TT is associated with shorter BMFS compared with 1L-therapy with ICI (CTLA-4+PD-1 or PD-1 alone), independent of other prognostically relevant factors. In addition, our multivariate analysis showed that 1L-therapy with CTLA-4+PD1 leads to better OS than 1L-therapy with PD-1, CTLA-4 or BRAF+MEK. In contrast, these correlations could not be demonstrated in the multivariate analysis for PFS, which showed only inferiority of CTLA-4 monotherapy compared with all other types of 1L-therapy. Also, in the multivariate analysis, we could not detect significant differences in survival outcomes of BRAFwt patients who received combined or single-agent PD-1 based ICI.

Our study shows a significantly higher incidence of MBM in BRAFmut patients who received BRAF+MEK as 1L-therapy compared with those patients who received first-line ICI. Melanoma is one of the cancers with the highest risk to develop brain metastases, accounting for approximately 6%–11% of all metastatic brain lesions.15 Several studies tried to predict the risk of development of MBM, but these comparisons were performed between patients who developed brain metastases in earlier stages independent of therapies.16–18 Frenard et al compared the development of MBM in a small cohort of 52 patients who received ipilimumab with a cohort of patients who received vemurafenib, and found no difference in risk of developing brain metastases between both cohorts.19 A French group analyzed 293 melanoma patients without brain metastases who were treated with either anti-PD-1 or anti-PD-L1 antibodies or other systemic therapies (including BRAF inhibitors and chemotherapy) and found a lower incidence of MBM in the PD-1 group compared with patients treated with other therapies.20 The results of this study are weakened by the fact, that BRAFmut patients were not assessed separately in the multivariate analysis and patients treated with BRAF+MEK were not compared directly to patients treated with PD-1. A retrospective study by Wang et al assessed BMFS in a cohort of BRAFmut, but not BRAFwt patients, receiving ICI or BRAF+MEK TT.21 This work showed a prolonged BMFS for patients who received ICI as 1L-therapy compared with patients who received BRAF+MEK. These findings were based on propensity scored matching of single variables and not on a multivariate Cox regression analysis including all important prognostic variables as in our analysis.

Why 1L-therapy with BRAF+MEK TT leads to a faster and more frequent development of MBM than ICI in BRAFmut patients is unclear yet. It is known, that while BRAF+MEK therapy shows good intracranial responses, the duration of these responses is relatively short.22 While there are data from a prospective phase II study on the effectiveness of dabrafenib+trametinib on MBM, such data is lacking for vemurafenib+cobimetinib or encorafenib+binimetinib. We did not detect differences in survival outcomes with the three different BRAF+MEK combinations in a subgroup analysis, but it has to be kept in mind, that the majority of our patients received dabrafenib+trametinib and that the other two groups were rather small.

Findings from different studies showed that MBM have distinct molecular features such as increased activation of the PI3K-AKT pathway and larger fractions of dysfunctional CD8+T cells with distinct expression of immune checkpoints compared with extracerebral metastases.23 24 This might explain why intracerebral effectiveness of BRAF+MEK is lower than outside the brain. In addition, Seifert et al25 showed in vitro that cerebrospinal fluid reduced cell death mediated by BRAF inhibitors. Interestingly, recently Wang et al demonstrated in a murine melanoma model that only a short sequence of PD-1 therapy followed by BRAF+MEK treatment was sufficient to suppress MBM development and improve the survival of the animals which was accompanied by T cell clonal expansion in intra- and extracranial metastases.26 These observations support our finding, that ICI as 1L-therapy are associated with better intracranial effectiveness and longer BMFS in BRAFmut patients. In the BRAFmut patients of our studied cohort, 1L-therapy with BRAF+MEK was associated with a significantly prolonged PFS compared with PD-1 and CTLA-4, but not to CTLA-4+PD-1. This observation is in line with the known higher primary resistance rate to PD-1 monotherapy compared with BRAF+MEK or CTLA-4+PD-1. The observed superior OS with CTLA-4+PD-1 compared with BRAF+MEK did not result from superior PFS of CTLA-4+PD1, which could be explained by a better response to second-line therapies in BRAFmut patients progressing on 1L-therapy with CTLA-4+PD-1, which is most often BRAF+MEK, compared with those progressing on 1L-therapy with BRAF+MEK, who mainly receive ICI as second-line treatment. Our findings for OS are in line with the recently published results of the randomized phase III trial Dreamseq.11 This study evaluated the sequential non-adjuvant therapy with ipilimumab+nivolumab followed by dabrafenib+trametinib in comparison to the converse sequence in BRAFmut melanoma. It has to be considered though, that the numbers of treated patients in the sequential arms of this study were very small. Nevertheless, an exploratory analysis of survival data from the most relevant clinical trials on BRAF+MEK TT and ICI therapy performed by us also led to similar results: a comparison of the mean PFS and OS data at 3 years after treatment start revealed a clear superiority of ICI vs TT as 1L-therapy (3 year OS 41.3% for BRAF+MEK, 49.9% for PD-1, and 58.4% for CTLA-4+PD-1).27 A similar trend was detected in the 3-arm randomized prospective SECOMBIT trial.10 Patients who received first-line therapy with encorafenib+binimetinib followed by second-line therapy with ipilimumab+nivolumab showed a slightly lower OS at 2 years (65%) compared with patients with the converse sequence (73%) in this trial. In addition, there was a so-called “sandwich”-arm with an 8 week run-in phase followed by ipilimumab+nivolumab until progression and then switch to TT. OS for this arm was in between the two other arms (69%). All differences were not statistically significant, but showed a clear trend. A recent update confirmed these findings at 37.1 months median follow-up.28 The real-world findings of our present study now underline these retrospectively obtained findings and those of the prospective Dreamseq and SECOMBIT trials. In addition, several studies have reported immunological changes in the tumor microenvironment after progression on TT, which can explain the inferior clinical activity of ICI as second-line therapy after progression to BRAF+MEK.29–31 In line with this and the results of our study, a recent study compared patients with MBM who received CTLA-4+PD-1 as first-line therapy or after progression on BRAF+MEK.32 There, treatment with CTLA-4+PD-1 after progression on BRAF+MEK showed a very low response rate which was associated with an enrichment in genes from the innate anti-PD-1 resistance signature (IPRES).

Another remarkable observation made in our present study is that a significant number of patients did not receive a second-line therapy. Further analyzing this interesting point, we found several reasons for this: (1) good response to 1L-therapy which does not make a second therapy necessary, (2) rapid tumor progression on 1L-therapy with fast deterioration of overall health status ruling out the start of a second-line therapy, (3) adverse events of 1L-therapy that make a second-line therapy impossible, or (4) patient wish. This finding of a high rate of patients never receiving second-line therapy underlines the importance of the right choice for an optimal 1L-therapy. When only considering 1L-therapy, in our study OS was significantly better in BRAFmut patients who received CTLA-4+PD-1 compared with all other treatment types. Moreover, this superiority was independent of other prognostically relevant factors such as age, ECOG-PS, serum LDH, and disease stage). While in BRAFmut patients OS on 1L-therapy with CTLA-4+PD-1 was superior compared with PD-1 alone, this difference was not detectable for BRAFwt patients. A similar finding was reported in the Checkmate-067 trial. In that study, the 6.5 years OS rates for BRAFwt patients were 46% on CTLA-4+PD-1 and 42% on PD-1, while in BRAFmut patients they were 57% on CTLA-4+PD-1 and 43% on PD-1.33 This separation of OS curves in BRAFmut patients treated with CTLA-4+PD-1 or PD-1 was already visible in the 4 years analysis.34 On one hand, this OS difference could be explained by the availability of BRAF+MEK TT as an efficient second-line therapy option in these patients, which BRAFwt patients do not have. On the other hand, different features in tumor biology and immunology intrinsic to these different types of melanoma might probably contribute to this effect, since BRAF activation is associated with the production of immunosuppressive cytokines, downregulation of MHC-class-I molecules, reduced T cell recognition, and a higher number of myeloid-derived suppressor cells and regulatory T cells, leading to a more immunosuppressive tumor microenvironment and a stronger immune escape of tumor cells in BRAFmut melanoma.35 36

Our study is not without limitations. Despite the large number of patients, the retrospective nature of this study has clear limitations: while we tried to account for many important prognostically relevant parameters, information on some parameters such as the sum and the biggest diameter of tumor burden and organ dysfunction by metastasis could not be included into the multivariate analysis, since they were not recorded in the registry. While LDH is one important marker of tumor burden, these parameters could have influenced treatment selection and therefore prognosis as well. Organ dysfunction or metastasis of critical organs, for example, could have prompted treating physicians to select treatment with TT instead of ICI, because of their faster treatment effect. In addition, in our real-world cohort, 50% of patients had elevated serum levels of LDH of whom more than half had very highly elevated levels (>2× ULN). This explains a significantly lower ORR to ICI and TT than reported for patients in most prospectively randomized trials, which has to be kept in mind, when results of such studies are compared with ours. It also has to be considered, that during the long time period over which the data was collected (2011–2021) not all therapeutic options were always similarly available. Particularly between 2011 and 2016, there were less approved therapeutic options available than nowadays. This could have influenced treatment outcomes and decisions toward first-line and second-line therapies. To partially account for this, we excluded patients who received BRAF monotherapy as 1L-therapy from this study, because the BRAF+MEK combination has become standard of care due to its better efficacy and tolerability. We also did not detect any significant differences for OS for the different time periods (before and after approval of the different therapies, online supplemental figure S1) other than for ipilimumab, which has not been used as first-line therapy any more since approval of PD-1 based ICI). In addition, a high number of missing values for the ECOG-PS reduced the number of patients that could be assessed in the multivariate analysis. Nevertheless, due to our high patient number, we could still analyze different subgroups and therapy sequences with robust statistical results. Further, the rather low incidence of MBM in the entire cohort (22.7%) has to be kept in mind, when interpreting the data for BMFS with different therapy sequences in the multivariate analysis. Because of much higher patient numbers for 1L-therapy than for second-line therapy, analyses of outcomes with 1L-therapy are suited better to show smaller differences between groups. Lastly, in our cohort, only a small number of patients received adjuvant therapy for stage III and IV disease, which can be explained by the long time period covered. Nowadays, most stage III patients receive adjuvant therapy with ICI or BRAF+MEK TT, which influences selection and effectiveness of first-line treatment in the non-adjuvant setting. This has to be kept in mind, when interpreting our data and drawing conclusions from it for treatment selection.

Altogether, our analysis of a large real-world cohort of melanoma patients demonstrates a faster and more frequent development of brain metastasis in BRAFmut patients treated first-line with BRAF+MEK TT, and prolonged OS in BRAFmut patients treated first-line with CTLA-4+PD-1 compared with PD-1 or BRAF+MEK. Moreover, we did not detect improved OS or BMFS in BRAFwt patients treated with combined ICI compared with PD-1 alone. These findings underline that survival outcomes have to be assessed separately for BRAFmut and BRAFwt patients, and that ICI, particularly CTLA-4+PD-1, should be preferably chosen as first-line therapy in BRAFmut melanoma patients without MBM.

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.

Ethics statements

Patient consent for publication

Ethics approval

Informed consent was obtained from all patients included into this study; the ADOREG registry was approved by the ethics committee of the University Duisburg-Essen (14-5921-BO).

Acknowledgments

We thank all investigators and patients participating in the ADOREG registry.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Correction notice Since this article first published, the authorship has been updated. Karin Scharffetter-Kochanek has been replaced by Anca Sindrilaru.

  • Contributors Conceptualization: CF and SU. Data collection: all authors. Analysis of data: CF and SU. Writing of manuscript: CF and SU. Editing and correction of manuscript: all authors. Guarantor: CF.

  • Funding CF was funded by the Köln Fortune Program of the University of Cologne. This research did otherwise not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

  • Competing interests All authors declare no conflicts of interest affecting this study. Conflicts of interest outside the submitted work are as following: CF has been on the advisory board or has received honoraria from Bristol Myers Squibb, Immunocore and Novartis and received travel grants from Bristol Myers Squibb, Novartis and Pierre Fabre. PM declares research support from Bristol Myers Squibb, Novartis and Merck Sharp & Dome; speakers and advisory board honoraria from Almirall Hermal, Beiersdorf, Bristol Myers Squibb, Merck Sharp & Dome, Immunocore, Merck Serono, Medac, Novartis, Pierre Fabre, Sanofi Genzyme, Sun Pharma and Roche, and travel support from Bristol Myers Squibb, Merck Sharp & Dohme, Novartis and Pierre Fabre. LB received honoraria from Amgen, Bristol-Myers Squibb and Sun Pharma. FriM has received travel support or/and speaker’s fees or/and advisor’s honoraria by Novartis, Roche, BMS, MSD and Pierre Fabre and research funding from Novartis and Roche. KK has served as consultant or/and has received honoraria from Amgen, Roche, Bristol Myers Squibb, Merck Sharp and Dohme, Pierre Fabre, and Novartis, and received travel support from Amgen, Merck Sharp and Dohme, Bristol Myers Squibb, Amgen, Pierre Fabre, Medac and Novartis. IG declares speakers and advisory board honoraria from Almirall Hermal, Bristol Myers Squibb, Merck Sharp & Dome, Novartis, Pierre Fabre, Sanofi Genzyme, Sun Pharma and Roche. RG: Invited speaker: Roche, BMS, MSD, Novartis, Amgen, Merck Serono, Almirall Hermal, SUN, Sanofi, Pierre-Fabre. Advisory board: BMS, Roche, Novartis, Almirall Hermal, MSD, Amgen, SUN, Sanofi, Pierre-Fabre, 4SC, Bayer, MerckSerono, Pfizer, Immunocore. Research grants: Novartis, Pfizer, Johnson & Johnson, Amgen, Merck-Serono, SUN Pharma, Sanofi. Travel/meeting support: Roche, BMS, SUN, Merck-Serono, Pierre-Fabre. JU is on the advisory board or has received travel support from: Amgen, BMS, GSK, Immunocore, Leo Pharma, MSD, Novartis, Pierre Fabre, Sanofi, Roche. JU has received research support from Novartis; speakers and advisory board honoraria. PT has been on the advisory board or has received honoraria from Almirall, Bristol-Myers Squibb, Novartis, Pierre-Fabre, Merck Serono, Sanofi, Roche, Kyowa Kirin, Biofrontera, and 4SC and received travel grants from Bristol Myers Squibb, and Pierre Fabre. RH reports speakers and advisory board honoraria from Bristol-Myers Squibb (BMS), Immunocore, Novartis, Pierre-Fabre, Roche and SUN pharma outside the submitted work. CP received honoraria (speaker honoraria or honoraria as a consultant) and travel support from: Novartis, BMS, Roche, Merck Serono, MSD, Celgene, AbbVie, SUNPHARMA, UCB, Allergy Therapeutics, Pierre Fabre, Kyowa Kirin and LEO. AF served as consultant to Roche, Novartis, MSD, BMS, Pierre-Fabre; received travel support from Roche, Novartis, BMS, Pierre-Fabre, received speaker fees from Roche, Novartis, BMS, MSD and CeGaT, outside the submitted work. She reports institutional research grants from BMS Stiftung Immunonkologie. FZ declares speakers and advisory board honoraria and/or travel support from BMS, MSD, Roche, Novartis, Pierre Fabre and Sanofi Aventis. FraM (Frank Meiss) served as a consultant and/or has received honoraria from Novartis, BMS, MSD, Pierre Fabre, Sanofi Genzyme, Sun Pharma and travel support from Novartis, Sun Pharma, Roche, Pierre Fabre and MSD. BS is on the advisory board or has received honoraria from Immunocore, Almirall, Pfizer, Sanofi, Novartis, Roche, BMS and MSD, research funding from Novartis and Pierre Fabre and travel support from Novartis, Roche, BMS and Pierre Fabre. AK reports receiving lecture fees and fees for serving on advisory boards from MSD Sharp & Dohme, Almirall, Infectopharm, and Boehringer Ingelheim. JW has been on the advisory board, received honoraria and/or travel grants from Bristol Myers Squibb, Novartis, MSD, and Pierre Fabre. TG has received speakers and/or advisory board honoraria and travel support from BMS, Sanofi-Genzyme, MSD, Novartis Pharma, Roche, Abbvie, Almirall, Janssen, Lilly, Pfizer, Pierre Fabre, Merck-Serono. CL has received speaker’s fees, advisory board honoraria and travel reimbursements from Merck, MSD, Roche, Almirall Hermal, Biontech, Sanofi, Sun Pharma, Kyowa Kirin, Immonocore, BMS, Pierre Fabre, Novartis. LH has received consultancy and speaker fees from Amgen, Biome Dx, BMS, Curevac, Merck, MSD, Myoncare, Novartis, Pierre-Fabre, Roche, Sanofi and SUN. SG has been on the advisory board and/or has received travel support from Bristol Myers Squibb, MSD, Sun Pharma and Novartis and received research support from Novartis and Pierre Fabre. DD has been on the advisory board or has received honoraria from BMS, Kyowa Kirin, MSD, Novartis, Pierre Fabre, Sanofi and received travel grants from Boehringer, Mylan, Pfizer. GS has received honoraria from BMS. GL has received travel support from Sun Pharma. LZ served as consultant and/or has received honoraria from Roche, Bristol-Myers Squibb (BMS), Merck Sharp & Dohme (MSD), Novartis, Pierre Fabre, Sanofi, and Sunpharma and travel support from MSD, BMS, Amgen, Pierre Fabre, Sunpharma, Sanofi and Novartis. EL served as consultant and/or has received honoraria from Amgen, Bristol-Myers Squibb, Merck Sharp & Dohme, Novartis, Medac, Sanofi, Sunpharma and travel support from Medac, Bristol-Myers Squibb, Pierre Fabre, Sunpharma and Novartis. JCB received speaker’s bureau honoraria from Amgen, Pfizer, Recordati and Sanofi, and is a paid consultant/advisory board/DSMB member for Almirall, Boehringer Ingelheim, InProTher, ICON, MerckSerono, Pfizer, 4SC, and Sanofi/Regeneron. His group received research grants from Bristol-Myers Squibb, Merck Serono, HTG, IQVIA, and Alcedis. DS declares relevant financial activities (Roche, Novartis, Bristol-Myers Squibb, Merck Sharp & Dohme, Sanofi, Regeneron, Array, Pierre Fabre, 4SC, Helsinn, Philogen, InFlarX, Merck-Serono, SunPharma, Ultimovacs, Sandoz). SU declares research support from Bristol Myers Squibb and Merck Serono; speakers and advisory board honoraria from Bristol Myers Squibb, Merck Sharp & Dohme, Merck Serono, Novartis and Roche, and travel support from Bristol Myers Squibb, Merck Sharp & Dohme, and Pierre Fabre.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.