One in three
What the MARS2 surgeons went on to say about their own trial

For two years I did not offer anyone an operation for mesothelioma. Nobody instructed me to stop. A randomised trial reported that the operation shortened lives, and I did what a surgeon ought to do with a result like that. What I did not expect was that the surgeons who ran the trial would publish a paper saying much the same thing I had been thinking — and then keep going.

What MARS2 found

MARS2 randomised patients with resectable pleural mesothelioma to chemotherapy alone, or to chemotherapy with extended pleurectomy/decortication. Median survival was 24.8 months without surgery and 19.3 months with it, and the difference in restricted mean survival at two years was −1.9 months. Serious adverse events ran at more than three times the rate in the surgical arm. The trial concluded that pleural mesothelioma should be regarded as unresectable, and practice across Britain changed within months. Mine included.

Guy's was one of the five surgical centres. This is not something that was done to us. We were part of it, and it found against an operation we performed.

Britain ran the experiments

It is worth saying plainly what the last fifteen years look like from here. MARS, in 2011, found excess deaths in the arm randomised to extrapleural pneumonectomy. MesoVATS, in 2014, found that video-assisted partial pleurectomy did not improve survival over talc pleurodesis, and cost more complications and more days in hospital to establish it. MARS2, in 2024, found against extended pleurectomy/decortication. Three randomised trials of surgery in mesothelioma; all British; all negative.

Very nearly the whole randomised evidence base for mesothelioma surgery was generated in this country. The centres that criticise those trials most loudly are, for the most part, the ones that never stopped operating and have case series to show for it. That record is not a British embarrassment. It is the reason a British surgeon's opinion on this is worth anything at all — and it is why, when MARS2 reported, the operation closed across the country in a matter of months.

What the trial's own surgeons said next

In February 2025 David Waller, writing as the lead recruiter in MARS2, published a paper with five of his co-investigators — one of them from Guy's — under a title that leaves little to interpretation: Why the MARS2 Trial Does Not Mean the End of All Mesothelioma Surgery.

The admission in it is unusually plain. They operated on too many patients with known poor prognostic factors. They operated too late in the course of the disease. And because the disease was advanced, they took too much tissue. One in seven patients recruited had non-epithelioid histology, a subtype every surgeon in the field already knew surgery could not outrun. PET-CT was not mandated. Mediastinal nodal status was never confirmed by biopsy. The selection criteria were deliberately liberal, because a trial that cannot recruit answers nothing — and the recruitment worked, ahead of schedule, through a pandemic. The price of that pragmatism became legible only afterwards.

One in three

Then they did the thing that gives the paper its weight. They took 79 of the 158 patients who had surgery — half the surgical arm, from four of the five centres — and restaged them against the ninth edition of the TNM classification, using the pleural-thickness measurements the trial had never made.

Twenty-seven of them, one in three, would have been offered an operation under contemporary criteria: stage I or II, epithelioid. The other two-thirds failed on grounds that do not look like fine judgement — non-epithelioid histology on the diagnostic biopsy, locally advanced disease, involved nodes that an endobronchial ultrasound would have found before anyone picked up a knife.

The median survival of that one in three was 32 months. Of the two-thirds who should not have been operated on: 8.5.

The diaphragm, and what surgery used to mean

Extended pleurectomy/decortication includes taking the diaphragm, and in MARS2 the diaphragm was resected in 83% of cases — including in patients whose pathology showed it had never been involved. Losing a hemidiaphragm costs a patient breathing they may not have to spare. Treatment-related mortality was 12% in the surgical arm against 6% without surgery, and it was driven mostly by pneumonia. It was not driven by the cancer: disease progression was no different between the two arms.

But the diaphragm is not the deepest problem. The deepest problem is what the operation was attached to. In 2014 I was one of the authors of a systematic review of pleurectomy/decortication within multimodality treatment, and what I took from that work then — and still think — is that the surgery was never the therapy. It was the part of the therapy that had to leave the patient well enough to receive the rest of it.

Set that against what MARS2 actually delivered: two cycles of chemotherapy, interrupted for an extensive operation, after which 40% of the surgical arm received no further systemic treatment of any kind. The programme I trained in looked different. Pleurectomy/decortication with hyperthermic pleural lavage; prophylactic radiotherapy; chemotherapy completed; and PET-CT surveillance afterwards, so that recurrence was found early enough for second-line treatment to be worth giving. We published the progression patterns from that surveillance in 2013. Whatever one concludes about its merits, it was a programme. MARS2 tested an operation dropped into the middle of a course of chemotherapy that, for four patients in ten, was never finished.

Where that leaves the patient in front of you

This year the Society of Thoracic Surgeons published an international expert consensus on the multimodal treatment of pleural mesothelioma. It does not overturn MARS2, and it does not try to. What it does is place surgery inside a multimodal regimen conditional on selection, on multidisciplinary assessment, and — pointedly — on the experience of the surgeon and the volume of the centre.

That is the argument, and it is uncomfortable for both sides. MARS2 is real evidence, the surgical arm did worse, and anyone who waves it away is not being serious. But a trial that operated on the wrong two-thirds tested something other than the question it is now used to close: whether an early, epithelioid, properly node-staged mesothelioma, in a patient who can complete the rest of the treatment, is better served by an operation or without one. That question is still open, and the honest answer is that we do not know.

For most patients who come to me with mesothelioma the answer remains no, and the work that matters is the biopsy, the staging and the control of the effusion. For a small minority it is not. Telling those two groups apart was never the trial's job. It is mine.

Related: Forty-eight hours, then call a surgeon · A nodule is not a diagnosis · Pleural disease

← Back to the Journal