Immunotherapy before surgery kept every dMMR colon cancer patient alive at five years

Giving two kinds of immunotherapy in the few weeks before surgery, rather than chemotherapy after it, kept every single patient with a particular form of colon cancer free of their disease at three years and alive at five, according to the long-term results of the NICHE-2 trial [s1]. The numbers are the kind that rarely appear in oncology — 100% on both counts — and they come from a cancer subtype whose biology makes it unusually vulnerable to the immune system [s1].

The subtype is mismatch-repair-deficient (dMMR) colon cancer, which accounts for roughly one in seven cases. These tumours have lost the cellular machinery that fixes DNA copying errors, so they accumulate mutations at a furious rate. That makes them look intensely foreign to the immune system — and exquisitely responsive to checkpoint inhibitors, drugs that release the brakes on immune cells [s1]. The logic of giving those drugs before surgery is that the tumour, still in place, acts as a target that trains the immune system broadly, potentially clearing microscopic cells that have already escaped and would otherwise seed a later relapse [s1].

What the trial did

NICHE-2 was a single-arm, multicentre, phase 2 trial run at six Dutch hospitals [s1][s2]. It enrolled patients with non-metastatic dMMR colon cancer and gave them a short, sharp course of immunotherapy before their planned operation: a single dose of ipilimumab at 1 mg per kg on day 1, plus two doses of nivolumab at 3 mg per kg on days 1 and 15 [s1]. Surgery followed within weeks. Of 123 patients screened, 115 were treated and 111 formed the per-protocol population [s1]. The trial's earlier read-out, reported in 2024, had already shown that this brief regimen wiped out the tumour in a large majority of patients at the time of surgery; this analysis asked the longer question of whether that translated into durable, relapse-free survival [s1].

What it found

It did, completely. The three-year disease-free survival — the trial's primary endpoint — was 100% (97.5% confidence interval 100 to 100), meaning not one patient had their cancer return within three years of surgery [s1]. An exploratory analysis of overall survival at five years was also 100% (95% CI 100 to 100) [s1]. The median follow-up was long enough to give those figures weight: 58.7 months (interquartile range 49.0 to 72.9) for disease-free survival and 60.4 months (IQR 50.6 to 74.1) for overall survival [s1].

The trial also tracked circulating tumour DNA (ctDNA) — fragments of cancer genetic material in the blood — as a window on how deeply the treatment worked. Among patients whose tumours showed a near-complete response, ctDNA cleared in 19 of 27 (70%); among those with a complete response, it cleared in 60 of 65 (92%) [s1].

How to read it

A 100% survival figure demands scepticism, and the trial's design supplies the caveats. NICHE-2 is single-arm — there was no control group given standard care, so the comparison is to historical expectations rather than a randomised placebo or chemotherapy arm [s1]. It is phase 2, modest in size, and conducted at a handful of specialist centres [s1]. Confidence intervals that read "100 to 100" reflect the fact that zero events were observed, not a guarantee that the true relapse rate is exactly zero; with more patients or longer follow-up, some events would be expected [s1].

Even with those limits, the result is striking because it fits the biology and because it is now backed by years of follow-up, not months [s1]. It also reframes the standard approach. Conventional practice removes the tumour first and considers chemotherapy afterwards. A separate, randomised trial recently showed that adding the checkpoint inhibitor atezolizumab to post-surgery chemotherapy raised relapse-free survival in the same dMMR subtype. NICHE-2 makes the more radical case that in these tumours a short burst of immunotherapy before surgery may do the heavy lifting, potentially sparing some patients chemotherapy altogether [s1].

What to watch

The open questions are about generalisability and de-escalation. Whether a six-centre Dutch phase 2 result holds in larger, more varied populations is the first; randomised trials testing neoadjuvant immunotherapy against standard care in dMMR colon cancer are the way that gets answered [s1]. The second, raised by the ctDNA data, is whether some patients respond so completely that surgery itself could eventually be scaled back — an organ-sparing question that this trial raises but does not resolve [s1].

This article describes research and is not medical advice. Cancer treatment decisions belong with a patient's oncology team.