What the margin can't reach
Three things that make me argue against a segmentectomy

Segmentectomy is selected on the size of the tumour, and then it succeeds or fails on things that size cannot see.

Two centimetres is a real boundary and I am not arguing with it. Both trials that made sublobar resection acceptable set the ceiling there, and there is settled agreement in thoracic surgery not to go beyond it for an intentional segmentectomy. What interests me is what happens once a tumour is through that gate. Inside the eligible group the local outcomes are not uniform, and the supplementary analysis of JCOG0802 published last year (Nakagawa et al, J Thorac Oncol 2025;20:157–166) sets out how uneven they are.

Inside the gate, the numbers separate

Locoregional relapse after segmentectomy in the trial was 8.9%, against 6.9% after lobectomy. That headline covers a wide spread. A pure-solid appearance on thin-section CT roughly tripled the odds of relapse. Tumours in the lingula relapsed in 23.5% of cases; tumours in the left upper division, in 5.5%. A margin narrower than the tumour was recorded in 18.5% of all segmentectomies, and in around a quarter of right basal and left S6 resections.

Some of that margin figure belongs to the laboratory rather than the operating theatre. The staple line comes off, the cut surface is inked, and several millimetres go with it. Every thoracic surgeon has had that conversation with a pathologist, and the trial's own data hints at it: among the patients with inadequate margins the macroscopic median was 17mm and the pathological median 10mm. It does not make the finding meaningless, but it does mean part of the number was created after the specimen left the room.

Three factors, three different responses

What struck me reading the paper is that the risk factors do not behave alike. The question worth asking of each one is whether it argues for a different operation, a different margin, or a different plan altogether.

A micropapillary component argues about the operation. Nitadori and colleagues (JNCI 2013;105:1212–20) found that in adenocarcinoma of 2cm or less, a micropapillary component of 5% or more raised five-year recurrence after limited resection from 12.4% to 34.2%, while after lobectomy the same threshold made no significant difference. A feature that only causes harm in the smaller operation is what you want to see before you let pathology change your plan. It is why micropapillary gives me pause about segmentectomy when other adverse features do not.

A pure-solid appearance argues about the margin, not the operation. In the same JCOG analysis pure-solid was associated with relapse after lobectomy as well. And the post-hoc analysis of the pure-solid subgroup (Hattori et al, Lancet Respir Med 2024;12:105–116) found overall survival was better after segmentectomy than after lobectomy in these patients, hazard ratio 0.641. Pure-solid marks a tumour that behaves worse wherever you cut. It makes me want a generous margin and attentive follow-up rather than a lobe.

Segment location argues about deliverability. Median margins in the trial were 30mm in the left upper division and lingula, and 20mm in left S6 and the right upper lobe. That is anatomy rather than biology — how much parenchyma sits between the tumour and the planes you have to divide — and it is visible on the CT before anyone scrubs. The question is live: the segment harbouring the tumour was the subject of correspondence in the same journal from the Memorial Sloan Kettering group (Caso and Rocco, J Thorac Oncol 2025;20:e23–e24), who have argued its prognostic value deserves more weight than it currently gets.

The relapses a good margin did not prevent

None of this is an argument that margin does not matter. A Copenhagen series of 155 consecutive thoracoscopic segmentectomies (Huang and Petersen, Ann Thorac Surg 2025;119:316–324) found margin distance an independent predictor of locoregional recurrence, with a breakpoint at 20mm arrived at from the data rather than inherited from protocol. Recurrence occurred in 14.2% at a median of 17.1 months. Two centimetres turns out to be a defensible number twice over.

Which makes the next finding harder, not easier. Eleven patients in the JCOG trial relapsed at the surgical margin. Seven of them had a margin wider than the tumour. The standard response — take more lung next time — would not have helped those seven.

The likeliest explanation is spread through air spaces, defined and reported by Kadota, Adusumilli, Travis and colleagues at Memorial Sloan Kettering in 2015 (J Thorac Oncol 2015;10:806–814), in a paper specifically about recurrence after limited resection in small stage I adenocarcinoma. JCOG0802 had closed to accrual in October 2014 and collected no data on it. So the largest randomised evidence we have on segmentectomy is silent on a mechanism described as the trial was closing, by the group whose earlier work gave us the micropapillary threshold.

Eguchi and colleagues (J Thorac Oncol 2019;14:87–98) later matched patients and found that in tumours positive for air-space spread, sublobar resection did worse than lobectomy whatever the margin-to-tumour ratio. In tumours without it, local recurrences clustered in those with a ratio below 1. Margin width mattered in one group and much less in the other.

That work is retrospective and propensity-matched, and the literature is not consistent — some series find air-space spread prognostic after sublobar resection only, others after lobectomy. It is a reason for caution rather than a settled mechanism. But it accounts for why widening a margin sometimes fails, and I have not come across a better explanation.

What we can know beforehand

A preoperative biopsy can report a micropapillary component, and when it does I take it seriously. What it cannot do is reassure. Huang and colleagues (J Thorac Cardiovasc Surg 2017) compared biopsy with the resected specimen in 128 patients and found sensitivity for a micropapillary or solid component of 16.5%. Even where that component made up 40% or more of the tumour, biopsy identified it in about a quarter of cases. The test is worth acting on when positive and close to silent when negative, and the two need reading differently.

Frozen section looks more promising for air-space spread — 71% sensitive and 92% specific in the Eguchi cohort — but it is rarely asked to look for it, and I include myself in that.

Where this leaves the decision

None of this makes segmentectomy the wrong operation. The trials stand, and the case for lung-sparing resection in the right tumour has not weakened. What the follow-up analyses suggest is that the criterion we select on — a size, with a radiological ratio attached — is doing its job as a gate and no more than that. Most of what determines how the operation goes sits behind it.

For me that means three things worth reconsidering rather than three rules. A micropapillary component reported before surgery is a reason to revisit the plan. A pure-solid tumour in a segment whose anatomy will not give a generous margin is a combination worth pausing over. And a wide margin should not be mistaken for having dealt with the problem, because for some tumours it never was the problem.

← Back to the Journal