Regarding the letter from Roxburgh,et al.

Dear Editors,

We appreciate Roxburgh and colleagues’ thoughtful response to our paper. ¹ We share the common understanding that the Multiple Sclerosis Severity Score (MSSS) is intended and best suited to characterize groups of patients rather than individuals. That it can easily be calculated for an individual patient has made it tempting for researchers and clinicians to use it as a surrogate marker for a patient’s expected disease trajectory.

In their letter, the authors state that in creating the MSSS they “made no assumptions of its stability” over time. However, we would argue that the very concept of devising such an algorithm for MS disease severity assumes that data from single time-point assessments “are representative of disease severity and not unduly influenced by sampling variation or disease fluctuation,” according to their original paper. ² They further state that “single point assessment data from as early as year 1 can be used to represent disease severity” and that over 15 years, “the mean change in MSSS was about zero.”

We sought to evaluate, in our cohort of 122 patients, the extent to which this is true in practice.

We found that the mean MSSS declined persistently to reach a nadir at year 7, seesawed somewhat through year 10, then started to increase through the 19th year of follow-up. On an individual basis, the MSSS frequently diverged from the original measurement, not simply going down after the baseline assessment, but often oscillating, and ultimately yielding disability trajectories either substantially better or worse than would have been predicted.

It is certainly possible that some assessments were made at times of relapse, as Roxburgh et al. suggest. In general, our clinical assessments were made on an annual basis, reflecting the real-world nature of this data set. If initial exacerbations and subsequent recoveries had a major impact on our findings, however, then we would not have expected the mean MSSS to remain roughly stable from year 1 to year 2, as depicted in Figure 4. The authors contend that the leftward pointing arrows in Figure 2, where we show dispersion from projected deciles, reflect recovery from relapse. Although this could be true to a small degree, those arrows depict disease course over many years, and relapse recovery is typically complete over a period of months. Although our Kaplan-Meier plot does show some early deviations from baseline MSSS at the one-year mark, the vast majority of such changes occur subsequently.

The authors correctly point out that our sample size was limited, particularly in the later years of follow-up. But the “wild fluctuations” we observed in MSSS over time are perhaps a more accurate representation of the unpredictability of this disease. What we sought to characterize was the degree of individual heterogeneity and variability of trajectory. Our modest sample size was sufficient to demonstrate this.

Even acknowledging our methodological limitations, our paper ultimately reinforces and further characterizes the cautionary note that the authors included in their original publication: “one should be cautious about using MSSS to predict future disease severity in any single patient: any such prediction will be subject to considerable uncertainty.”

As our ability to treat both relapsing and progressive forms of MS continues to increase, the disease trajectory that patients can anticipate will likely only deviate more favorably from that which was calculated from cohorts in decades past. This is in no way a poor reflection on the MSSS, but rather a measure of progress in our field since it was originally devised.