Variable 2 r

Placebo p-value IFN-p-lb r p-value r Overall p-value Year 1 relapse rate Year 1 Tj lesions 0.305 < 0.0001 0.120 0.1280 0.264 < 0.0001

Year 2 relapse rate Year 2 lesions 0.198 0.0145 0.319 0.0014 0.278 < 0.0001

Year 3 relapse rate Year 3 Tj lesions 0.254 0.0031 0.238 0.0022 0.251 < 0.0001

Year 0-3relapse rate Year 0-3 Tg lesions 0.240 < 0.0001 0.203 0.0002 0.249 < 0.0001

Year 1 EDSS change Year 1 Tj lesions 0.130 0.0095 -0.040 0.5366 0.078 0.0483

Year 2 EDSS change Year 2 Tj lesions 0.141 0.0100 0.121 0.0961 0.113 0.0101

Year 3 EDSS change Year 3 Tj lesions 0.211 0.0005 0.124 0.1226 0.184 0.0002

Year 0-3 EDSS change Year 0-3 lesions 0.208 < 0.0001 0.118 0.0147 0.184 <0.0001

9,6 Discussion

This is the first placebo-controlled trial to report the effect of a IFN-p-lb preparation on the evolving pathological process in secondary progressive MS. It is also by far the largest MS study to date in which MR analysis has been performed. It reveals that IFN-p-lb has a substantial effect on the pathological evolution of disease in patients with secondary progressive MS, and that this effect was sustained for the duration of the trial (up to 3 years). This treatment effect was readily apparent both for the total volume of brain pathology visible on the Tj weighted images, and for the numbers of active Tj lesions appearing during the study period. Furthermore, statistically significant, though generally modest, correlations were observed between a number of MRI and clinical measures of disease activity and progression. The data strongly support the clinical evidence for efficacy demonstrated in this trial (European Study Group on IFN-p-lb in SPMS 1998). The results are now discussed in more detail, first considering the MRI findings per se and secondly, the clinical/MRI correlations.

9.6.1 MRI findings

The baseline T2 lesion volume was similar and generally large in both treatment arms of the

study; mean and median Tj lesion volume values were approximately twice those reported in previous trials involving cohorts with relapsing remitting disease. This is consistent with the results presented in Chapter 6 (where analysis of a multicentre database revealed very similar

differences between the MS subgroups) and elsewhere (Gawne-Cain et al, 1998). The enormous

range of individual T2 lesion volumes is of course also evident (Table 9.4), as is the modest

correlation of baseline T2 lesion volume with EDSS (r = 0.09, p = 0.002), confirming that, for

individual patients with secondary progressive MS, the T2 lesion volume is not a reliable indicator

In the placebo group, median Tj lesion volume increased significantly year upon year. The overall median percentage increases from baseline (1.6%, 2.4% and 11% at 1,2 and 3 years respectively) are somewhat lower than the 5-10% per annum changes typically reported in relapsing remitting

MS placebo cohorts (Paty et al, 1993; IFNB MS Study Group 1995; PRISMS 1998; Simon et

al, 1998). This largely reflects the much higher baseline T2 lesion volume in the secondary

progressive cohort, and the absolute increase in T2 lesion volume is similar to that found in early

relapsing remitting cohorts. A more or less linear increase in T2 lesion volume from year to year

might have been anticipated, but in fact a greater increase was seen in the third year (Table 9.6). Between years 2 and 3, several sites changed or upgraded their MRI scanner, with an increase in field strength and a noticeable improvement in image quality. Such changes can certainly have

an important impact on measured T2 lesion volume (Filippi et al, 1997a), and this probably

explains the year 2-3 result. However, any such effect would apply equally to both treatment arms (due to the randomisation schedule, treatments were balanced within centres), and the difference in outcome between the treatment groups was still clearly evident: thus during year 3 the median

change in T2 lesion volume was an increase of 1.05 cm^ in the placebo group and 0.37 cm^ in the

IFN-p-lb group (p = 0.0002).

In the treated group, there was a significant and substantial reduction in T2 lesion volume seen

at the first follow up. The same phenomenon has been reported in two previous studies in

relapsing remitting MS (Paty et al, 1993; PRISMS et al, 1998). This is likely to reflect both; (i)

the striking resolution on T2 weighted images that is often seen in inflammatory/oedematous

lesions of recent origin, even without therapeutic intervention, and (ii) the effect of treatment on inhibiting new lesion development. It is also possible that IFN-p-lb treatment facilitates a more rapid and complete resolution of such active lesions, by suppressing inflammation and creating

a more favourable environment for effective repair, although, as yet there is no evidence for this. In the IFN-P-lb group, the effect of resolution of existing inflammatory/oedematous lesions clearly outweighed the increase in Tj lesion volume due to the small number of new lesions that

were identified despite treatment. However, this reduction in T2 lesion volume was not seen

beyond the first annual follow up, which indicates a more transient mechanism, most likely the natural resolution of pre-existing inflammatory lesions.

The treatment effect on the annual assessment of T2 lesion activity was significant and sustained

throughout the study. IFN-P-lb treatment was associated with a 57% reduction in the mean and 80% reduction in the median number of new/enlarging lesions compared to placebo over the duration of the study. A similar level of treatment effect was apparent in each of the three years of the study and this effect is at least as powerful as that previously demonstrated in relapsing

remitting MS cohorts (Paty et al, 1993; Simon et al, 1998; PRISMS Study Group 1998). This

suggests that the inhibition of new lesion development with IFN-P-lb treatment is common to both relapsing remitting and secondary progressive MS subgroups.

The correlation between T2 lesion volume and T2 lesion activity in the placebo group was

significant but modest (r = 0.36 over the whole study period). This is not perhaps surprising,

given that the net change in T2 lesion volume represents a composite of increase due to new lesion

formation, together with decrease due to lesion resolution. Furthermore, the impact of new lesions

on the change in T2 lesion volume is not linear over time, with activity shortly before the exit scan

having a greater impact than earlier activity (see Chapter 6). A recent study has also confirmed

that the net change in T2 lesion volume substantially underestimates volume of new T2 lesions that

effect, together with the inevitable attenuation of correlations due to measurement error (both in qualitative and quantitative analysis), probably accounts for the only modest overall correlation

between the changes in T2 lesion volume and T 2 lesion activity. It is also notable that the

correlation between these outcomes for the whole study period is weaker in the IFN-P-lb group (r = 0.15, p = 0.001), and this was most apparent when the first 12 months of change is considered (r = 0.09, p = 0.067); this dissociation is likely to reflect the initial treatment effect of reducing

T2 lesion volume in the face of ongoing (albeit diminished) new lesion activity.

When the relationship between baseline T2 lesion volume and subsequent T2 lesion volume

change was studied, an interesting difference was found between the placebo and treated groups

(Table 9.12). In the placebo group, there was a positive correlation between baseline T2 lesion

volume and both absolute change in T2 lesion volume over the next 3 years (r = 0.18, p < 0.0001)

and the cumulative number of new/enlarging lesions (r = 0.18, p < 0.0001). This may be a

reflection of the fact that those patients with a larger T2 lesion volume at study entry had

accumulated abnormality more rapidly than those with a lower T2 lesion volume prior to study

entry, and this correlation may therefore merely be a manifestation of this effect continuing on into the study.

In contrast, in the IFN-p-lb group there was a negative correlation between baseline T2 lesion

volume and subsequent T2 lesion volume change (r = -0.16, p = 0.0003). It is likely that patients

with higher T2 lesion volumes will also have a higher load of inflammatory/oedematous lesions

with the potential to resolve, and that this accounts for the greater decrease in T2 lesion volume,

9.6.2 Clinical/MRI correlations

A significant, albeit modest, correlation was found between the change in T2 lesion volume and

EDSS over the study period and this was equally apparent in both the placebo and IFN-p-lb groups (Table 9.13). Thus, treatment did not result in a dissociation such that the profound MRI related treatment effect was no longer concordant with the clinical benefit. Slightly stronger

correlations were apparent between the annual T2 lesion activity and EDSS change. These

correlations support the use of such MRI measurements as clinically relevant markers of disease evolution in secondary progressive MS. Notwithstanding, it is also apparent that the magnitude of the treatment effect on MRI and clinical outcomes is quantitatively different (the more pronounced effects being seen on MRI), and the modest degree of clinical/MRI correlation

suggests that other pathological processes than those studied using T2 weighted brain imaging

contribute to the evolution of disability. The major factors likely to contribute to this clinical/MRI dissociation have been discussed in Chapter 1.

The nature of the EDSS scale also may contribute to the limited relationship between this

parameter and T2 lesion volume. It is noticeable that in a year by year cross-sectional analysis of

the correlation between T2 lesion volume and EDSS, the level correlation increased with each year

of follow up (at baseline: r = 0.09, p = 0.002; at 3 year follow up: r = 0.17, p < 0.0001). The initial EDSS range at entry was relatively narrow - from 3 to 6.5 - but increased during follow up as some, but not all, patients accrued more disabilities. As has been previously noted (Gawne-Cain

et al, 1998), a greater range of disabilities as indicated by a widened range of EDSS

measurements provides a better opportunity for appreciating the MRI-EDSS relationship.

extent of pre-existing irreversible pathology; areas of chronic persistent demyelination might be an unsuitable environment for maintaining axonal viability; axonal degeneration may eventually occur even in the absence of an ongoing active pathological process. In this study, there was a

weak overall correlation between baseline T2 lesion volume and EDSS change over the next 3

years (r = 0.067, p = 0.020), implying that T2 lesion volume is modestly predictive of subsequent

clinical progression in established MS. This suggests that existing disease load in secondary progressive MS may contribute, to a minor extent, to future clinical evolution. The situation is

somewhat different earlier in the disease course, where T2 weighted imaging can be powerfully

predictive of clinical evolution (Filippi et al, 1994; O’Riordan et al, 1998). This discrepancy

supports the concept that pathological changes other than those defined by T2 weighted imaging

become increasingly important contributors to clinical progression later in the disease course.

Significant correlations were also identified between relapse rate and T2 lesion activity for all

annual timepoints and over the whole study period (Table 9.14). This is not perhaps surprising, given; (i) the well known relationship between gadolinium enhancing lesion activity and relapse rate, and (ii) the fact that a substantial proportion of gadolinium enhanced lesions are also seen

on corresponding T2 weighted images. It is clear, however, that a monthly gadolinium enhanced

imaging protocol is more sensitive to short term disease activity than monthly T2 weighted

imaging (Miller et al, 1993). Furthermore, given the natural resolution of new T2 lesions over a

period of weeks to leave a small residual abnormality, an additional proportion of new T2 lesions

will not be visible with an inter-scan interval of 12 months. Therefore, visual analysis of annual

T2 weighted imaging has a limited role as a marker of short term disease activity. However, the

results suggest an important role as a marker of disease progression. The fact that treatment effect

with clinical/MRI correlations that are least as strong as for Tj lesion volume quantification, raises a question of the need for Tj lesion volume analysis in such a study. The logistic requirements of electronic data acquisition, transfer of electronic data to a coordinating centre, and subsequent quantitative analysis are enormous. In contrast, simple visual assessment on hard copy is vastly less logistically demanding, requiring only 10-15 minutes per pair of MRI studies. Given that the most important criteria of any surrogate are objectivity, efficiency and clinical predictive value, it could readily be argued that qualitative annual Tj activity analysis is a more appropriate outcome measure than Tj lesion volume quantification. This is further supported by the MRI

results of the Avonex IFN-P-la study (Simon et al, 1998), which found a powerful treatment

effect on Tj lesion activity (supporting the clinical outcome data), without any significant effect

on T2 lesion volume.

In conclusion, the MRI results of this study support and supplement the clinical efficacy endpoint measures. Highly significant and powerful treatment effects have been identified for both qualitative and quantitative analysis of the annual MRI studies. The results also suggest that the existence of significant clinical/MRI correlations is common to both secondary progressive and

relapsing remitting MS, extending previous observations on relapsing remitting cohorts (Paty et

al, 1993; IFNB MS Study Group 1995) to the later secondary progressive phase of the disease

course. These results may also guide future phase III trial design, by highlighting the utility of simple visual assessment of Tj weighted MR studies, and suggesting a greater role for this outcome in future phase III studies. However, the modest nature of the clinical/MRI correlations confirm the appropriate role of clinical endpoints as primacy efficacy variables, and highlight the need for newer MRI tools.