Treatment optimization and combining metrics
NEDA – new approach
NeuroSens Survey: How do you assess treatment response in practice?
The proliferation of disease-modifying therapies (DMT) in multiple sclerosis has focused attention on the need for improved measures for evaluating treatment efficacy and response. In part, this is driven by the increasing difficulty of detecting a treatment effect in clinical trials of early MS patients with minimal disease activity, as well as the need to differentiate DMTs with respect to their varying effects on clinical and radiological endpoints. In clinical practice, identifying suboptimal responders is a matter of some urgency since therapies are more likely to provide benefit in the first 2-5 years after diagnosis of CIS/MS. An early treatment effect may be the only effect, so it is important to optimize the drug regimen before the window of opportunity for effective intervention is shuttered.
One challenge is the limitations of individual measures of disease activity. Both early relapses and MRI activity are prognostic of disability progression in the first two years after diagnosis, but have limited predictive value thereafter (Weinshenker et al. Brain 1989;112(part 1):133-146; Fisniku et al. Brain 2008;131:808). The EDSS scale, if used in practice, is insensitive to small changes in neurological or neuropsychological functioning, most notably in patients with minimal impairment.
Once treatment has been initiated, MRI measures, such as T2 lesion number or volume, are predictive of disability progression even in the absence of relapses or EDSS change, at least in interferon-treated patients (Rio et al. Mult Scler 2009;15:848-853). However, this was not the case when the Barcelona group analysed data for patients on glatiramer acetate: isolated MRI activity during the first 12 months of treatment was not associated with a higher risk of new disease activity or progression (Rio et al. Mult Scler 2014;20:1602-1608). This may be due to the pharmacodynamics of glatiramer acetate or the differing modes of action of these agents. Similar analyses have not been performed with the newer DMTs.
Treatment optimization and combining metrics
Several strategies have been proposed to improve the assessment of treatment efficacy and response. One approach has been to use individual metrics alone or in parallel to provide an evidence basis to clinical gestalt (Freedman et al. Can J Neurol Sci 2013;40:307-323). Thus, the clinician’s level of concern when assessing treatment response may be characterized as Low, Medium or High in each of the domains of relapses, EDSS progression and MRI, with one High, two Mediums or three Lows suggestive of a suboptimal response. A somewhat more aggressive strategy, with one Medium/High in any domain indicating a possible need to switch therapy, has also been proposed (Correale et al. J Neurol Sci 2014;339:196-206).
A second approach is to combine metrics. An aggregate measure of relapses + MRI (modified Rio score) has been shown to be highly predictive of disease progression (Sormani et al. Mult Scler 2013;19:605-612; Fahrbach et al. BMC Neurol 2013;13:180, free full text at www.ncbi.nlm.nih.gov/pmc/articles/PMC4225567/pdf/1471-2377-13-180.pdf). The next step was to add EDSS change to determine the proportion of patients that achieved disease activity-free (DAF) status, defined as no relapses, no sustained disability progression and no MRI activity (no Gd+ lesions, no new/enlarging T2 lesions) (Havrdova et al. Lancet Neurol 2009;8:254-260). DAF was sometimes mischaracterized as “disease-free”, and subsequently acquired the designation of “no evidence of disease activity” (NEDA).
To date, two-year NEDA rates have been reported for natalizumab (37% vs. placebo 7%) (Havrova 2009); fingolimod (33% vs. placebo 13%) (Kappos et al. AAN 2011; abstract PD6.002); dimethyl fumarate 240 mg BID (28% vs. placebo 15%) (Giovannoni et al. AAN 2012; abstract PD5); and teriflunomide (23% vs. placebo 14%) (Freedman et al. AAN 2012; abstract PD5.007). NEDA rates in two-year trials using active comparators have also been reported for DMF and glatiramer acetate (18% vs. 12%) (Havrdova. AAN 2013; abstract P07); and alemtuzumab versus subcutaneous interferon-beta-1a (39% vs. 27%) in de novo patients (Cohen et al. Lancet 2012;380: 1819-1828) and previously treated patients (32% vs. 14%) (Coles et al. Lancet 2012;380:1829-1839). Results are not directly comparable due to differences in study populations, how NEDA is defined, and incomplete data.
While achieving NEDA has been described as “intuitively appealing”, this benchmark has a number of limitations (Bevan & Cree. JAMA Neurol 2014;71:269-270). Individual measures, such as relapses and EDSS progression, have not been standardized and different analyses may employ different criteria. In the post-hoc analysis of the CombiRx trial, NEDA rates were considerably different depending on how broadly relapses were defined (Lublin et al. Ann Neurol 2013;73:327-340). For the combined interferon/glatiramer acetate cohort, the three-year NEDA rate was 33.3% using protocol-defined relapses, and 26.9% using non-protocol-defined relapses (rates using ‘suspected relapses’ were not reported). In practice, NEDA results will depend on how frequently a patient is evaluated and the range of assessments that are performed. ‘No evidence’ of activity does not imply no activity, nor that evidence does not exist.
To date, only one study has examined the value of NEDA in predicting long-term disability outcomes. In a seven-year follow-up of patients in the Comprehensive Longitudinal Investigation of MS at the Brigham and Women’s Hospital (CLIMB) study, the positive predictive value (PPV) of NEDA for no significant progression (defined as EDSS change < 0.5) was 71.7% at one year, and 78.3% at two years (Rotstein et al. ECTRIMS 2014; abstract P763). The PPV was improved somewhat when disability was assessed with the Timed 25-Foot Walk (T25FW) rather than EDSS.
While some may argue that NEDA simply formalizes what clinicians already do in practice, it has focused attention on whether current assessments are adequate and sufficiently sensitive to detect suboptimal treatment response in a timely manner. A multifactorial MS decision model using somewhat different assessments was recently proposed (Stangel et al. JAMA Neurol 2014;71:1056-1057), and a full paper has now been published (Stangel et al. Ther Adv Neurol Disord 2015;8:3-13). The model is essentially an admixture of NEDA and the treatment optimization recommendations (TOR; Freedman 2013). Decision-making is guided by traffic lights rather than speedometers, perhaps to shift focus from the clinician’s concern to the broader road ahead. The recommended four domains to be evaluated are relapses, progression, neuropsychology and MRI.
The most significant difference is that the model replaces EDSS with a battery of tests (T25FW, 9-Hole Peg Test, Symbol Digit Modalities Test, Low Contrast Sloan Letter Chart). The value of MRI findings is downgraded so that lesion activity alone is not sufficient for a change in therapy, in contrast to the TOR model (Freedman 2013). A treatment switch may be prompted, however, by significant worsening of MS fatigue, as assessed by the Fatigue Scale for Motor and Cognitive Functions (FSMC; Penner et al. Mult Scler 2009;15:1509-1517). Fatigue scores in the model may be offset by coexisting depression or anxiety (which subtract points). Thus, the proposed model advocates that a more complete assessment should include fatigue and cognitive function – key aspects of patient quality of life – although it is unclear if switching therapies will benefit these areas.
A different approach is to include endpoints that more closely reflect the underlying pathophysiology of MS. As noted by one of the authors of the multifactorial decision model, changes in brain volume measures (e.g. rate of ventricular enlargement) appear to be more strongly predictive of disease progression than lesion measures (Lukas et al. J Neurol Neurosurg Psychiatry 2010;81:1351-1356). In the short term, stratifying patients according to normalized brain volume at baseline (High, Medium and Low) has been shown to be predictive of disability progression at two-year follow-up (Sormani et al. ACTRIMS-ECTRIMS 2014; abstract PS12.3). The MAGNIMS group has also reported that whole-brain and central atrophy are predictive of EDSS at 10 years (Popescu et al. J Neurol Neurosurg Psychiatry 2013;84:1082-1091). A subsequent meta-analysis of 13 treatment trials found that the impact of DMTs on the rate of brain volume loss in the first two years of therapy was correlated with disability progression (Sormani et al. Ann Neurol 2014;75:43-49); this correlation was independent of the effect of therapies on active lesions.
These studies suggest that brain atrophy is a potentially modifiable risk factor for disability progression, which may provide a rationale for the use of higher efficacy agents in the first 24 months of therapy. One assessment of treatment efficacy using NEDA + brain atrophy (NEDA-4) has been performed to date using data from two placebo-controlled trials of fingolimod (Kappos et al. ECTRIMS 2014; abstract FC1.5). For this analysis, NEDA was defined as no relapses, no EDSS progression confirmed at six months, and no new/enlarged T2 lesions; Gd+ lesions were not considered since they added no value once T2 lesions had been considered. Brain volume loss occurs in healthy individuals, so a cut-off value for percentage brain volume change (PBVC) of -0.4%/year was proposed. This roughly corresponds to twice the normal rate of brain volume loss seen in age-matched non-MS individuals (Barkhof et al. Nat Rev Neurol 2009;5:256-266).
At two years, the proportion of patients who achieved NEDA-4 was 19.7% with fingolimod versus 5.3% with placebo (Kappos 2014). Between-group differences were significant (odds ratio 4.41) in favour of fingolimod. Similar odds ratios were seen when other cut-off values for annual PBVC were used, from -0.2%/year (OR 4.0) to -1.2%/year (OR 4.25). The proportion of patients with brain volume loss below the cut-off value was significantly higher with fingolimod versus placebo (37.2% vs. 26.7%) (De Stefano et al. ECTRIMS 2014; abstract P290). An interesting observation was that the treatment effect was greater in the subset of patients with lower versus higher normalized brain volume at baseline (41% vs. 11% treatment effect) (Sormani et al. ECTRIMS 2014; abstract PS12.3), suggesting an added benefit of therapy in patients at highest risk of disability progression.
NEDA-4 uniquely provides some degree of assessment of both the inflammatory and neurodegenerative components of MS, and represents the most stringent criteria thus far proposed for evaluating treatment response. While brain atrophy is not routinely evaluated in clinical practice, these preliminary findings suggest that the use of DMTs that slow brain volume loss early in the disease course may result in clinically meaningful reductions in disability progression over the longer term.
NeuroSens Survey: How do you assess treatment response in practice?
This poll is currently disabled.