Complete data were available on 689 participants, of whom 586 (85%) were female with classic RTT, 79 (11%) with atypical RTT and 24 (4%) without clinical features of RTT. A MECP2 mutation was identified in 555/586 (95%) in classic RTT, 53/79 (67%) in atypical RTT and 24/24 (100%) in participants without clinical features of RTT. For the entire cohort, scoliosis was noted in 292/586 (50%) with classic RTT, 31/79 (39%) with atypical RTT and 5/24 (21%) in participants without clinical features of RTT. Focusing on the 586 classic RTT participants, we analyzed the prevalence of scoliosis with MECP2 mutation type and clinical parameters collected at enrollment. As shown in , 546/586 (93%) participants with classic RTT had a detectable MECP2 mutation including 348 (59%) with one of the eight most common mutations (R106W, R133C, T158M, R168X, R255X, R270X, R294X and R306C), 37 (6.3%) with 3’ frameshift deletions, 50 (8.5%) with large deletions, and 111 (19%) with other mutations. In addition, 554/586 (95%) participants reported whether or not they had scoliosis at enrollment; 292/554 (53%) were confirmed to have scoliosis.
Mutation distribution and scoliosis occurrence in 586 participants with classic RTT
The mean age for classic RTT participants was 10 years ranging from <1 to 57 years; 38% were under 5 years and 28% were greater than 13 years; this older age group comprised 45% of participants with scoliosis and only 9% of those without scoliosis. The mean age was 15 years for those with scoliosis and 6 years for those lacking scoliosis participants, producing a significant between-group age difference (p <0.0001 using two sample t-test). In as much as nearly 40% of the cohort were < age 5 years; this frequency distribution is not surprising. For those participants age 16 years or older, 85% reported scoliosis.
We examined overall clinical severity scores on both the CSS and MBA, BMI z-scores using CDC standardization, absent or lost ability to sit or walk, hand use, seizures, bone fractures, constipation, peripheral vasoconstriction, age at regression, and head growth and compared these with the presence or absence of scoliosis. As age or mutation type could confound the relationship between scoliosis and each clinical parameter, two models, one adjusting for age alone, the other adjusting for both age and mutation type were used to investigate the association. As shown in , the increased prevalence of scoliosis was significantly associated with higher overall clinical severity scores on both the CSS and MBA, delayed acquisition, loss, or absence of sitting and walking, occurrence of seizures, constipation, peripheral vasoconstriction, and poor head growth after adjusting for age and mutation type. In our cohort, no participant with scoliosis reported acquired and maintained hand use, while 6% of those without scoliosis had preservation of acquired and conserved hand use. Including all significant clinical parameters in the model, we constructed the multiple logistic regression model with backward variable selection. In this model, severity score on MBA, later acquisition, loss, or absence of walking, and occurrence of constipation retained the positive association with scoliosis ().
Relationship between scoliosis and clinical parameters: univariate.
Significant clinical parameters from multiple logistic regression model for P (having scoliosis) = 1 adjusting for age and mutation type (backward variable selection)
Examination of specific mutations showed that R133C, R294X and R306C are significantly associated with decreased risk of having scoliosis compared with all other mutations after adjusting for age (), but only for R294X and R306C after additionally adjusting for the three significant clinical parameters: MBA, walking, and constipation. However, note that R133C has lower frequency than R306C (4% vs. 7%) but higher prevalence of scoliosis (32% vs. 29%) in this sample. Among participants with one of the eight common mutation types, we further examined the effect of decreasing risk of scoliosis in R294X. Participants with T158M, R168X, R255X, and R270X had significantly higher risk of developing scoliosis than those with R294X in both models ().
Relationship between common mutation types and scoliosis
Relationship between 8 common mutation types and scoliosis
Scoliosis surgery was performed in 70/554 (13%) participants reporting whether or not they had scoliosis and was less likely to have been done in individuals with R133C, R294X, and R306C mutations. The mean age (± standard deviation) at surgery was 12 years (±3 years), and the median was also 12 years (range 5–22 years). Individuals requiring surgery were much less likely to have acquired walking or to have retained previously acquired walking, namely, 65/71 (92%) compared to 6/71 (8%) who had acquired walking prior to surgery and retained walking afterwards.