A cohort of 192 patients with facial PWS was seen between March 2011 and January 2013. The mean age at the time of analysis was 8·3 years (SEM 0·4) and the mean patient follow-up period was 8 years. Seventy-five patients had an ophthalmological examination, of whom 55 were diagnosed with glaucoma. One hundred and twenty-one patients had an MRI of the brain, of which 90 were abnormal. Forty-nine of 162 patients for whom there were reliable data had seizures, and 59 of 143 had neurodevelopmental delay. As a result of these investigations 104 children were diagnosed with SWS, using the definition of any facial PWS plus either MRI abnormalities or glaucoma. The frequencies of the original classification of V1–3 are shown in Table 1.
Strikingly, there was no significant association between the trigeminal nerve distribution and an abnormal MRI. Associations between the eight facial areas from the second classification and abnormal MRI were then tested, and independently significant areas were found to be the central forehead and lateral forehead. Upper-eyelid involvement and bilateral distribution were initially significant but became nonsignificant when combined with the lateral forehead and central forehead. We therefore defined a new area that was the best predictor of an abnormal MRI, which equates to the 'forehead'. This is delineated at the lateral and inferior margins by a line joining the outer canthus of the eye to the top of the ear, and including the upper eyelid. This area covers parts of the distribution of all three branches of the trigeminal nerve (Fig. 1b).
Once the forehead area was identified as the relevant area this was used to model the association with clinical outcome measures. Of the 103 children in this cohort with involvement of the forehead, 83 had SWS, and 20 had only the facial PWS. However, this figure cannot be extrapolated to the total population as our cohort is selected for more severely affected patients. For patients with any involvement of this area the odds ratio of neurodevelopmental abnormality was 24·7 [95% confidence interval (CI) 3·2–188·8, P = 0·002], of seizures 15·8 (95% CI 2·1–120·5, P = 0·008) and of glaucoma 14·4 (95% CI 1·8–113·3, P = 0·011). Absolute numbers of those in each group are shown in Table 2, with Fisher's exact P-values. Bilateral involvement was not independently significant in any of these models when an interaction variable was used. On closer examination of the data we found that only six cases had bilateral involvement that did not include the forehead on either side. None of these six had any clinical or radiological adverse outcome.
Interestingly, when abnormal MRI was included in the model for clinical outcome measures, this was a better predictor of all clinical outcome measures than the forehead distribution of PWS, with odds ratios for seizures of 80·3 (95% CI 9·0–714·6, P < 0·001), neurodevelopmental abnormalities 24·7 (95% CI 3·2–188·8, P = 0·002) and glaucoma 14·4 (95% CI 1·8–113·3, P = 0·011). However, guidelines based on clinical phenotyping are proposed for practical purposes so that MRI can be targeted to the highest-risk infants.
The British Journal of Dermatology. 2014;171(4):861-867. © 2014 Blackwell Publishing