Prognostication of metastatic death in uveal melanoma patients: A Markov multi-state model
Introduction
Hepatic metastases are the primary cause of death in patients with uveal melanoma; however, tumour dissemination is only rarely detectable at the time of primary ocular treatment. There is a need for prognostic tools to estimate the risk of metastatic death and to predict when this might happen. If sufficiently reliable, such tools would enable medical care to be personalized, so that patients with a low risk of metastasis can be reassured while targeting special measures, such as counselling and systemic surveillance, at those who are likely to succumb to their disease. Since many patients with uveal melanoma are elderly, estimation of time to metastasis helps to predict whether death is likely to be caused by their uveal melanoma or by unrelated disease(s).
We previously developed a prognostic tool, the Liverpool Uveal Melanoma Prognosticator Online (LUMPO) model, that estimates all-cause mortality [1]; however, such an endpoint is not ideal for the following reasons: the cause of death is not usually difficult to ascertain; death from unrelated disease or age is common; and treatment or disease-related factors do not increase the risk of death from other causes [2].
The aim of the present study, therefore, was to develop a prognostic model of metastatic death.
Section snippets
The data
The model was developed with data from 4161 patients treated for uveal melanoma at the Liverpool Ocular Oncology Centre. Patients were included in the study if they resided in England, Scotland or Wales, and if their tumour involved the choroid. Diagnosis was based on clinical findings and, if these were inconclusive, on morphological examination of a biopsy. Tumour location and intraocular spread were determined by ophthalmoscopy and slit-lamp examination. Tumour dimensions were measured by
Results
Fig. 1 shows the cumulative incidence functions of metastatic death versus death from other causes. The curves show that the probability of metastatic death exceeds the probability of death from unrelated disease up to about 18 years post treatment; from this point onward, the latter predominates.
In Table 3, we report the averaged coefficients (over the ten fitted models) for the two causes of death, with Wald statistics p-values and hazard ratios (and attendant 95% confidence intervals.) The
Discussion
The C-index and the calibration curves indicate that our model provides reliable, personalised metastatic death prognostication.
The main strengths of our study are the large number of patients, the abundant genetic and histological data, the long follow-up and the accurate reporting mortality. The main weakness is the missing histological and genetic data in a significant proportion of patients. This is because methods for genetic analysis of small tumours samples and techniques for biopsy of
Conflicts of interest
None declared.
Acknowledgments
Funding: This work was supported by The Eye Tumour Research Fund of the Royal Liverpool University Hospital Trust.
The authors also thank the anonymous referees for their comments, which helped improve the presentation of the manuscript.
References (14)
- et al.
Estimating prognosis for survival after treatment of choroidal melanoma
Prog. Retin. Eye Res.
(2011) - et al.
A comparison of cause-specific melanoma mortality and all-cause mortality in survival analyses after radiation treatment for uveal melanoma
Ophthalmology
(1998) - et al.
Patterns of BAP1 protein expression provide insights into prognostic significance and the biology of uveal melanoma
J. Pathol.: Clin. Res.
(2018) - et al.
MRI in the detection of hepatic metastases from high-risk uveal melanoma: a prospective study in 188 patients
Br. J. Ophthalmol.
(2013) - et al.
Modeling Survival Data: Extending the Cox Model
(2000) - et al.
Mstate: Data Preparation, Estimation and Prediction in Multi-State Models. Version 0.2.10 [software]
(2016 Dec 4) - et al.
Survival and Event History Analysis: a Process Point of View
(2008)
Cited by (34)
Metastatic uveal melanoma: The final frontier
2022, Progress in Retinal and Eye ResearchCitation Excerpt :The metastatic rate of 50% is reached in 10 years in studies using the Kaplan-Meier analysis, which ignore competing causes of death, and in 25 years if competing causes of death are taken into account (Kujala et al., 2003). Therefore, the former studies exaggerate the real-life risk of metastasis-related death (Fig. 6) (Bergman et al., 2003; Eleuteri et al., 2018; Gooley et al., 2001; Kujala et al., 2003). Kaplan-Meier analysis and Cox regression are designed to presume that uveal melanoma would be the only possible cause of death, because their basic assumption is that the risk of dying of metastasis is the same before and after censoring, but obviously, a patient cannot die of metastases when they were censored for already having died of other causes (Aalen et al., 2008).
Intraocular and Orbital Cancers
2022, Comprehensive PharmacologyPRAME Expression: A Target for Cancer Immunotherapy and a Prognostic Factor in Uveal Melanoma
2023, Investigative Ophthalmology and Visual Science