Twelve reasons delays in uveal melanoma treatment might affect prognosis

The idea that a prolonged interval between diagnosing and treating uveal melanoma can affect prognosis has been at the centre of an active debate.1,5 This issue focuses on a key question for aggressive cancers that spread early: Does treating the primary tumour per se improve survival, or is it merely palliative once micrometastases have formed?^{6 7}

While local therapy cannot remove tumour cells that have already spread beyond the eye, starting treatment earlier may reduce additional metastases by stopping further growth or new mutations.

(1) In a recent single-centre retrospective study of 1145 consecutively diagnosed patients with posterior uveal melanoma, those treated ≥1 month after diagnosis (mean interval: 57 days) had a significantly higher competing risk incidence of metastatic death in AJCC stages II and III, but not in stage I, compared with patients treated within 1 month (mean interval: 20 days).^{1 4} Given that patients treated ≥1 month from diagnosis had smaller tumours and less frequent BAP-1 protein loss in the available cases, it is unlikely that they had a higher proportion of aggressive genetic traits. A more plausible explanation is that tumour growth during the interval between diagnosis and treatment resulted in a more advanced stage at the time of treatment. This interpretation is further supported by data from 12 cases in which tumour volume was measured at both diagnosis and treatment; after a mean interval of 390 days, the mean primary tumour volume increased from 355 mm³ to 769 mm³, with an estimated doubling time of approximately 368 days based on an exponential growth model.

(2) This doubling time aligns with our recent systematic review and meta-analysis of 199 choroidal melanomas and 87 growing nevi from five studies, which showed that uveal melanomas double in volume approximately every 360 days.⁸ Specifically, doubling times were approximately 717, 421 and 307 days for small, medium and large melanomas, respectively, whereas growing nevi took around 6392 days to double. The data imply that choroidal melanomas grow in a manner consistent with super-exponential kinetics—ie, bigger tumours accelerate their doubling rate, increasing the survival impact of treatment delays. A mixed-effects model further indicated that each month of deferred treatment for small, medium or large melanomas raised the 10 year risk of metastatic death by 0.3, 1.8 and 4.0 percentage points, respectively. These observations, rooted in independent datasets and multiple statistical approaches, reinforce the original conclusion.

(3) An independent cohort in a submitted (not yet published) study likewise found that intervals beyond ~30–45 days between diagnosis and treatment were significantly associated with worse disease-specific and overall survival.⁹ Using unmatched and propensity-matched Kaplan–Meier analyses and competing risk regression, we confirm that the delay-survival association holds true regardless of whether the day of diagnosis or the day of treatment is used as the baseline. This also suggests that time-to-treatment initiation bias did not substantially influence the results.

(4) The 360-day doubling time for all melanomas, and even the 717-day doubling time for small melanomas, identified in our meta-analysis is notably shorter than the 4.3% annual diameter growth (equating to a volume doubling time of about 2018 days, assuming a semi-spheroidal shape and similar growth in thickness) reported by Damato et al.² Since Damato’s study assumes a slower growth rate, direct comparisons are not straightforward. However, when the growth rate is closer to the observations in our meta-analysis, we arrive at broadly similar estimates of how treatment delays affect metastatic risk: Damato et al observed that a 20% annual increase in tumour diameter (around a 464 day volume doubling time) raises the absolute risk of metastatic death by 5.0 percentage points (pp) per year (approximately 0.4 pp monthly) for a 61-year-old female with a small melanoma (9.8 mm basal diameter, 2.9 mm thickness) harbouring monosomy 3, compared with 0.6 pp yearly (about 0.05 pp monthly) for the same-sized tumour with disomy 3. Table 1 summarises the estimates of the prognostic implications per month of treatment deferral across the four studies.

Table 1. Estimates of the prognostic implications of a 1 month treatment deferral from three studies.

Study	Data source	N	Key statistical methods	Result: influence on prognosis per month delay between diagnosis and treatment*
Stålhammar 20241	Single institution, retrospective.	1145	Competing risk regression; cumulative incidence estimates from competing risk data.	A 1% increase in the hazard of metastatic mortality (adjusted for LBD and tumour thickness). Significantly higher incidence of metastatic death in AJCC stages II and III, but not in stage I, for patients treated ≥1 month versus <1 month from diagnosis.
Damato 20242	Single institution, retrospective; survival estimates referenced from LUMPO.	24	Estimation of tumour growth rates using a linear mixed‐effects model; absolute risk of 15‐year metastatic mortality.	For small tumours (eg, a 60‐year‐old female with an 8.6 mm LBD tumour and 2.4 mm thickness): absolute increase of approximately 0.4 percentage points if monosomy 3 and 0.05 percentage points if disomy 3.
Stålhammar 20258	Systematic review and meta‐analysis; primary tumour growth rates from five studies across three countries; reference data from two independent sources including LUMPO.	199 melanomas and 87 growing nevi.	Linear mixed‐effects model to estimate the increase in 10‐year competing risk incidence (10CRI) of death from metastatic uveal melanoma. Estimation of effective exponential increase rates (EI10CRI).	Absolute increase in 10 CRI per month delay: 0.3 percentage points for small tumours, 1.8 percentage points for medium tumours and 4.0 percentage points for large tumours. Effective exponential increase rates (EI10CRI): approximately 0.2 percentage points for small tumours, 2.4 percentage points for medium tumours and 6.8 percentage points for large tumours. Using LUMPO, the average increases were 0.1, 0.8 and 4.1 percentage points for small, medium and large tumours, respectively.
Moghadam 20259	Single institution, retrospective.	336	Unmatched and propensity-matched Kaplan–Meier analyses and competing risk regression.	Worse overall survival for those treated >30 days from diagnosis; worse disease-specific survival for those treated >45, >60 or >90 days from diagnosis. Absolute increase in 10 CRI per month delay: 0.03 percentage points for small tumours, 0.8 percentage points for medium tumours and 2.1 percentage points for large tumours.

^*Using similar growth rates in the Damato (2024) study as reported in the meta‐analysis by Stålhammar (2025).

AJCC, American Joint Committee on Cancer; LBD, largest basal diameter.

(5) In a related meta-analysis by our team of seven case reports and case series, we examined 17 patients with untreated choroidal or ciliary body melanomas that were initially small- to medium-sized and not treated for over 5 years following the diagnosis. In this cohort—with some missing data and long, although generically reported, follow-up periods—all tumours with available follow-up data demonstrated growth, and the estimated 15-year Kaplan–Meier disease-specific survival was only 18%. Given the small cohort size and the retrospective nature of the included studies, these survival estimates must be interpreted with caution. Nonetheless, these data support the notion that the natural course of uveal melanomas leads to metastasis in a very high proportion of cases, underscoring the potential impact of timely treatment.¹⁰

(6) Aggressive traits such as monosomy 3, gain of 8q and epithelioid cytomorphology are present in a subset of the smallest tumours and become more common as tumour size increases. In our recent report, approximately 15% of ciliary body and choroidal melanomas with an estimated volume of 25 mm³ exhibited monosomy 3, compared with over 70% in tumours with a volume of 1800 mm³. Similarly, gains of chromosome 8q were observed in about 8% of tumours at 25 mm³ vs 75% at 1800 mm³, and epithelioid cytomorphology was present in 33% vs 74% of tumours at these volumes, respectively.¹¹ These findings indicate that although not all small tumours lacking these features will acquire them later, a subset does, thereby contributing to a more aggressive clinical course as tumour size increases.

(7) Literature on punctuated evolution indicates that canonical changes like BAP1 mutations often appear early, though further mutations can still arise later and drive metastatic progression.¹² Shain et al described how certain tumours developed additional driver mutations such as CDKN2A mutations, chromosome 3 loss and gains of 8q and 1p in the advanced phases, with those same aberrations present in metastatic deposits.¹³ In their cohort, about 14% (5 of 35) of cases of metastases emerged from later branches in the evolutionary tree. Of these, two cases (A50 and A60) involved small, likely asymptomatic, tumours, limiting the feasibility of earlier diagnosis and intervention. This highlights two key points: first, effective treatment of the primary tumour in the three remaining cases with larger tumours (representing 9% of the 35) before these additional mutations developed might have prevented the analysed metastases; second, if tumour cells had already disseminated from earlier tumour regions and established macrometastases, these likely carried fewer driver mutations, potentially resulting in slower growth.

(8) An international collaboration supports the association between larger primary tumours and more rapidly growing metastases. The study found that primary tumour size is significantly associated with survival in metastatic disease and the metastatic lesion burden at detection.¹⁴ A total of 128 patients from the Netherlands and 205 patients from Sweden with stage IV disease were included. In both cohorts, patients with large primary tumours (LBD ≥16 mm) or a higher American Joint Committee on Cancer stage at the time of initial diagnosis presented with more hepatic metastases and exhibited shorter overall survival from the time of metastatic detection, as estimated by Kaplan–Meier analysis. These findings suggest that primary tumour size may influence survival in metastatic uveal melanoma and that primary tumour treatment could provide a survival benefit by preventing further tumour growth and the accumulation of aggressive traits. However, caution is warranted in interpreting these results, as they diverge from previous studies.

(9) Straatsma and colleagues reported an unadjusted 5-year risk ratio for death of 1.79 (95% CI, 1.08 to 2.97) when comparing 42 patients from a natural history study (NHS) to 1,317 patients in the COMS medium melanoma trial.¹⁵ NHS patients were older but had smaller tumours at baseline. Of the 42 NHS patients, 22 eventually opted for treatment, with a mean delay of 1.4 years after declining initial participation. When adjusting for baseline patient age and tumour size, the risk ratio decreased to 1.54 (95% CI, 0.93 to 2.56), which was non-significant. However, no risk ratios were provided for time points beyond 5 years, despite additional mortality events occurring—seemingly at a higher rate in the NHS cohort. Moreover, although Kaplan–Meier survival curves were generated, no test statistics or P values were reported to assess the significance of survival differences between the NHS and COMS cohorts. Consequently, we conducted a re-analysis and found that the Kaplan–Meier overall survival estimate 8 years after enrolment was approximately 29% for COMS patients, compared with 45% for NHS patients (log-rank p=0.008)¹⁶

(10) Brachytherapy for posterior uveal melanoma using a 15 mm rather than a 20 mm plaque—without confirmation of plaque positioning—is associated with a greater incidence of local recurrence, with the hazard increasing as the difference between the plaque and tumour diameters decreases.¹⁷ Such patients also have a higher rate of metastatic mortality. No difference in local recurrence incidence between brachytherapy with 10 and 15 mm ruthenium plaques has been observed when plaque placements are verified using indentation or transillumination with a vitrectomy probe at each plaque margin; nonetheless, eccentric plaque placement was associated with local recurrence, with an HR of 3.4.¹⁸ Debate persists on whether local recurrence initiates metastasis or merely reflects a more inherently aggressive tumour phenotype. However, the absence of an overall increase in metastatic mortality in the 15 mm cohort suggests that local recurrence may indeed allow the tumour to acquire additional malignant features and continue seeding. In support of this, Bagger et al found that tumours that recurred after plaque therapy did not exhibit a greater prevalence of chromosome 3 or 8q abnormalities at baseline, reinforcing the notion that extended tumour growth can increase metastatic capacity.¹⁹

(11) Even though some may still advocate observation for small choroidal melanomas, a recent questionnaire indicates that a majority of experienced ocular oncology colleagues will now opt for treatment as long as a sufficient number of risk factors are present—indicating that the importance of timely treatment is no longer merely the opinion of a minority in our field.²⁰

(12) The risk of worsened prognosis due to treatment delay and tumour growth is not unique to uveal melanoma. Research involving over a million patients across various cancer types has demonstrated that even a 4 week delay in treatment can significantly increase mortality risk.²¹

It is not suggested that every suspicious lesion must be treated immediately, especially when the diagnosis is uncertain. Observation for growth can be crucial for confirming malignancy, and older or frail patients might not benefit from aggressive interventions. Nonetheless, these observations challenge the idea that, for most patients, primary tumour therapy is purely palliative and can be delayed.