Half of RAS-positive thyroid nodules were cancer — but the cancers rarely came back

Thyroid nodules are extremely common, most are harmless, and the central problem in managing them is telling the few dangerous ones apart from the many that are not. Molecular tests that scan a biopsy for mutations were meant to help — but a mutation in the RAS gene family has been an awkward in-between result, common in both benign and cancerous nodules. A large new series in The Journal of Clinical Endocrinology & Metabolism quantifies just how weak a cancer signal RAS is on its own, and suggests many such nodules could be treated less aggressively [s1].

What the study looked at

Researchers at a single institution reviewed a consecutive series of thyroid nodules that tested positive for a RAS mutation using the ThyroSeq molecular panel on fine-needle biopsy, and that had a known outcome — either surgical pathology or at least a year of follow-up ultrasound, over 2018 to 2024 [s1]. In all, 585 RAS-positive nodules were removed and examined, alongside 55 that were not resected [s1].

Under current practice, finding a RAS mutation on a biopsy typically pushes patients toward surgery, because the result cannot by itself rule cancer out [s1]. Longstanding American Thyroid Association guidance frames how nodules are stratified by ultrasound and biopsy and how far surgery should go when cancer is found or suspected [s2]. This study asks a sharper question: among nodules already known to carry RAS, what actually predicts cancer, and how do those cancers behave over time?

The distinction matters because the stakes of getting it wrong run in both directions. Treating every RAS-positive nodule as a probable cancer means operations that, in hindsight, were unnecessary for the many that turn out benign or indolent. Treating them too casually risks missing the minority that are genuinely aggressive. A test that only flags the presence of a mutation, without saying how dangerous it is, leaves that trade-off unresolved.

What the numbers showed

After surgery, the 585 nodules split roughly into thirds: 140 (24%) were benign, 149 (25%) were a borderline, indolent category called NIFTP — non-invasive follicular thyroid neoplasm with papillary-like nuclear features — and 296 (51%) were cancer [s1]. In other words, a bare RAS result was close to a coin toss.

The type of RAS mutation mattered. Nodules with an isolated KRAS mutation carried a cancer-or-NIFTP rate of 34%, well below the 74% seen with isolated NRAS and 81% with isolated HRAS [s1]. On multivariable analysis, three features flagged higher risk: extra chromosomal changes known as copy number alterations (odds ratio 12), larger nodule size on ultrasound (odds ratio 1.3 per unit) and a more worrying biopsy category [s1].

The behaviour of the cancers was the most reassuring part. Over a median 2.5 years of follow-up, persistent or recurrent disease occurred in just 19 of 272 patients with a cancer diagnosis (7%) [s1]. That risk was concentrated almost entirely in cancers that also carried additional high-risk mutations: 50% of those (14 of 28) recurred, against 2% (5 of 244) without them [s1]. Among cancers driven by an isolated RAS mutation alone, the recurrence rate was 0.8% — one patient in 133 [s1]. And of the 55 nodules left in place rather than removed, only 6 (11%) grew significantly over a median 2.3 years [s1].

What it means — and the caveats

The authors' conclusion is that molecular and clinical factors, taken together, predict cancer far better than a RAS result alone, and that isolated-RAS tumours behave so indolently they "may be managed with lobectomy, and in select cases, even non-surgically" [s1]. Removing half the thyroid rather than all of it, or watching rather than cutting, would spare some patients lifelong thyroid-hormone replacement and the risks of a bigger operation.

Several limits temper how far this should travel. It is a retrospective series from one centre, so practice patterns and the specific molecular panel used may not generalise, and nodules were included only if they reached surgery or had follow-up — a selection that can skew the mix. The follow-up, while informative, is still measured in a few years, shorter than the slow timescale over which some thyroid cancers recur. And "may be managed with lobectomy" is a hypothesis for guideline committees and prospective study, not yet a standard of care [s1].

For patients, the practical message is not to self-interpret a mutation report but to ask an endocrinologist what their specific RAS result — and any accompanying genetic findings — means for the extent of surgery. The broader significance is a shift in direction: molecular testing in the thyroid is maturing from a yes/no cancer flag into a tool for calibrating how aggressively to treat.