Premium
Tumour regression pattern according to tumour shape after brachyther‐apy for 330 uveal melanomas
Author(s) -
MAMUNUR R,
KIVELÄ T
Publication year - 2014
Publication title -
acta ophthalmologica
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.534
H-Index - 87
eISSN - 1755-3768
pISSN - 1755-375X
DOI - 10.1111/j.1755-3768.2014.4443.x
Subject(s) - regression , medicine , regression analysis , melanoma , ophthalmology , mathematics , statistics , cancer research
Purpose To describe regression patterns and their relationship with tumour shape of uveal melanoma by thickness and cross‐sectional area. Methods A retrospective study of 330 consecutive ciliochoroidal melanomas treated with I or Ru plaques in 2000‐2008 and imaged with Innovative Imaging I3‐ultrasound. Tumours were measured with Olympus DP‐Soft from original digital images. The COMS shape groups were partially collapsed: group 1: flat and crescent; group 2: oval and dome; group 3: mushroom; group 4: lobulated; and group 5: other or undetermined. Regression patterns were adapted from Abramson et al. (1990) into 4 main types: D (net decrease), S (stable), I (net increase), and M (for miscellaneous). We further subdivided M into 5 subtypes: DI (D>I), DS (D>S), ID (I>D), SD (S>D), and Z (zigzag; no trend). Results The regression pattern D was always the most common (31‐63% of tumors by thickness and 33‐72% by cross‐sectional area, for shape groups 1‐4); and the DS pattern was the second most common one (22‐24% and 16‐26%, respectively), except for group 2. Of the DI and Z patterns with later tumor increase, the former was fairly evenly spread across groups (5‐6% and 8‐10%, respectively) whereas the latter was observed mainly in shape groups 1 and 2 (15% and 17‐22%, respectively). The SD and ID patterns later tumor decrease were most common in shape groups 1 and 2, respectively, and most tumors representing the S pattern were shape group 1. Conclusion Heterogeneity in regression patterns and tumour shapes must be taken into account when using regression as a statistical variable in outcome analysis. Until now, all tumours have been pooled together despite significant differences in regression patterns, except in the study by Abramson et al. (1990).