PROCEEDINGS
`
`ADDRESSING THE CHALLENGES EN RISK ASSESSMENT
`
`AND RISK MANAGEMENT IN MULTIPLE SCLEROSIS
`
`Elliot M. Frohman. MD. PhD, FAANf Benjamin M. Greenberg, MD, MHSE john Ratchrord. Mai
`and Robert Zivadinov, MD. PhD. FAAN.§
`
`herapy for multiple sclerosis (MS) has
`undergone multiple evolutions in the last
`20 years. From the advent of the first US
`Food and Drug Administration (FD/’0‘
`approved therapy to the routine use of 4
`different injectahle medications, the field has come a
`long way in a relatively short period of time. With the
`re—release of natalizumah in 2006, patients with MS
`and treating physicians were faced with a new chal-
`lenge in disease therapymmore intensive risk/benefit
`discussions. After experiencing a “honeymoon“ period
`relative to the low risk associated with interferon and
`
`glariramer acetate injection therapy, patients with MS
`and physicians were forced to determine what amount
`of risk they wouEd be wiEling to endure in order to
`achieve substantially optimized disease-modifying
`effects (both clinical and radiographic).
`Correspondingly similar challenges are increasing
`with the emergence of novel therapeutic capabilities,
`based on targeting mechanisms not heretofore charac-
`terized in medical immunobioiogy. The coupling of
`greater treatment efficacy with the observation of a
`broader diversity ofassociared adverse events (some of
`which can be life threatening), will no doubt prompt
`the FDA and similar agencies around the world to for-
`
`
`
`*Proiessor ol Neurology and Ophthalmology, Kenneyfv‘lorie
`Dixon Pickens Disiinguished i’rolessor oi MS Research,
`Irene
`‘Nodel and Robert Alho Dislinguished Choir
`in Neurology.
`Director, MS Program and Clinical Center for MS, Universily of
`Texas Souihweslem Medical Center, Dallas, Texas.
`iAssisioni Professor oi Neurology, Siteclor, Transverse Myeliiis
`and Neuromyelrlis Opiico Program, Deputy Director, MS Center,
`Universiiy oi Texas Souihweslern Medical Center, Doilas. Texas.
`Ir’-\ssisioni Professor of Neuroiogy, Johns Hopkins Universriy
`School oi Medicine, Baltimore, Maryland.
`§Ditec£or, Buiiolo Neurormoging Analysis Cenier, Professor
`of Neurology, Jacobs Neurological
`lnsliiute, Schooi oi
`Bromedicoi Sciences, Slole University of New Yorit oi Buffalo,
`Buiicrlo, New York.
`
`muiare more complex approval processes, which will
`likely resuEt
`in a more protracted period of time
`between completion of phase Ill efficacy studies and
`the ultimate registration of these long-awaited
`advances on behalf of our deserving patients. After the
`recognition of potential
`life-threatening events with
`significant immunomodulation in patients with MS
`{in particular with progressive multifocai
`leukoen—
`cephalopathy [PML]J, both practitioners and patients
`have become more risk aware when deciding amongst
`therapies. Yet, the ongoing risk oflife-airering disabilw
`ity caused by MS persists, and these complexities
`weigh heavily on patients, families, and physicians.
`On October 1, 2011, a meeting was convened in
`Philadeiphia, PA, with approximately 50 academic
`and community-based neurologists who care for
`patients with MS, and who were appointed to serve as
`facuity in order to address the above-mentioned chal-
`ienges that face the neurologist responsible for provid»
`ing disease—modifying treatment for the patient with
`MS, commensurate with the intensity of the disease
`process, while taking into account the known risks of
`each of the considered treatments. The group consid—
`ered data that would help clinicians risk stratify
`patients relative to their disease course and severity.
`The group considered whether there are features of a
`patient at diagnosis, or early in the course of the dis—
`ease, that conld be utilized to prognosticate about the
`risk of filElll’C disability. The participants also reviewed
`data about the currently available FDA—approved ther-
`apies and their reported efficacy and risks. Ultimately,
`participants worked through a series of real—world
`patient vignettes in order to pracrically operationalize
`the available evidence—based data and expert Opinion
`for the purpose ofillustrating how specific clinical cir«
`cumstances can be translated into the rendering ofspo
`ciftc and rational treatment recommendations.
`
`A salient theme that was underscored by the par—
`
`
`bl. 12, No. 1 l March 2012
` 6 Page 1 of 12
`Biogen Exhibit 2218
`IPR 2018-01403
`
`Mylan V. Biogen
`
`Page 1 of 12
`
`Biogen Exhibit 2218
`Mylan v. Biogen
`IPR 2018-01403
`
`

`

`PROCEEDENG‘S
`
`
`ticipants throughout the meeting was that all therapeu-
`tic recommendations in the MS arena must be individu—
`
`alized. There are no amount of population data that will
`allow us to move patients through “cookie cutter” clini—
`cal management algorithms. Some patients are wiiling to
`take on more risk than orhers in order to optimize the
`chance of treatment-exacted remission and a diseaseofree
`
`status (at least by virtue of how we measure disease aetiv»
`ity; eg, no attacks, progression, lesions, etc). The goals of
`treatment are quite heterogeneous, contingent upon
`what is of greatest priority to the patient. For instance,
`some patients are more concerned with preservation of
`cognitive capabilities relative to physicai fimctioning.
`\Vhile the goals of disease—niodifi’ing therapy are princi—
`pally focused on reducing and mitigating inflammatory
`demyelinating attacks and disease progression, confusion
`often arises when patients erroneously assume that such
`treatments are also intended to eradicate their existing
`MS—reiated symptoms such as fatigue, cognitive slowing,
`heat and exercise intolerance, or bladder and bowel dys—
`function. Although these symptoms are likely a deriva—
`tive of the disease process, they are already established
`and thereby require a separate process of multidiscipli—
`nary symptom management. Other factors that influ—
`ence the type and intensity of treatment include patients
`who are considering pregnancy, while for others, adher—
`ence behavior wiil figure prominently in the uitimate
`choice of treatment. Collectively, these many variables
`are part of the complex decision—making process that
`both care providers and patients must confront: one that
`clearly corroborates the principle that there is definitely
`nor a one—treatment—fits-all approach.
`Notwithstanding the compelling need to personalize
`MS therapeutics, potentially usehil tools could be devel-
`oped to give patients and practitioners a way to assess the
`risk of the disease, the potential benefit of a therapy, and
`the relative risk of a serious adverse event while using a
`particular therapy. With a rapidly increasing therapeutic
`landseape for MS, a clinically practical “navigation” tool
`aimed at the application of “reasonable and safe” treat»
`ments that are tailored to each patient would represent a
`powerfiil advance for the clinical neurologist. lt is with
`this primary thrusr in mind that we organized the
`Philadelphia meeting, and upon which the framework of
`this monograph is based.
`
`tions, punctuated by variable lengths of remission. Most
`clinical trials for potential MS therapeutics have focused
`on reductions in annualized relapse rate (ARR) as the
`primary outcome measure of treatment efficacy. This
`outcome measure compares the AR for the placebo
`arm of a trial to the ARR of the treated patient cohort.
`In facr, no matter how effective a disease—modifying
`therapy may be, unless there is active worsening in the
`placebo group of a randomized, controlled clinicai trial,
`a therapeutic advantage cannor be established (eg, a
`false-negative or type il error). Alternately, if the place-
`bo group exhibits worsening in excess of what would
`represent typical MS disease activity, the active treat-
`ment may appear to be erroneously more effecrive (eg, a
`false—positive or type I error).
`While relapses cause disruption to patients’ lives,
`lost time from worlt, and hardships for families, there
`has been a vigorous debate about their effect on the
`overali course of the disease. Aiternativeiy Stated, do
`relapses matter? The evidence systematically reviewed
`as part of the MS Think Tank meeting suggests that
`they do in fact matter greatiy—particularly in the case
`of individual patients. First, data analyzed from the
`placebo arms of various randomized, controlled trials
`confirm that a significant number of patients have sus-
`tained accrued disability foilowing MS exacerbations.
`in Lublin’s carefully crafted and systematic examina»
`tion of the impact of MS attacks on compromised
`functional capabilities across a broad and representa-
`tive range ofciinical investigations, in excess onSO/u of
`patients will have a sustained l~point change on the
`Expanded Disability Status Scale (EDSS) following a
`confirmed relapse (Figure).’ Second, relapses represent
`ongoing disease activity not controiled by a therapy,
`and hence constitute a marker of ongoing disease pro«
`gression. Beyond suppression of relapses, however,
`
`Figure. Impact oi Relapses in Multiple Sclerosis
`
`Effects of Attacks on Disease Progression
`- N = 224 patients with 2% exacerbation
`— 90 days after exacerbation
`O 4 I96 had EDSS score rendual deficit of 20.5
`0 40% had EDSS score reSIdual deficit ofzi
`
`MS CLINICAL OUTCOMES
`
`- Attacks can lead to permanent worsening
`
`The most common form ofMS is relapsing—remit-
`ting MS (RRMS}, characrerized by acute exacerbaa
`
`5085 = Expanded Dlsability Status Scale.
`Data from Lublin et al.
`
`
`
`feline llopltins Advanced Slutlics in hletlicine I
`
`>-.l
`
`Page 2 of 12
`
`Page 2 of 12
`
`

`

`PROCEEDINGS
`
`
`there are many goals in MS therapy, the most impor-
`tant of which is the maintenance of both physical and
`intellectual function in our patients over decades.
`Multiple sclerosis is a heterogeneous disease of the
`central nervous system that has the potential to cause
`significant disability. Classically, the disabiiity of MS has
`been measured using the EDSS, which rates patients 0
`to 10 based on physical examination findings, and most
`particularly ability to walk. Although the EDSS score
`provides a metric For assessing disability in trials, it has
`several important limitations. First, the EDSS is heavily
`biased toward the physical domain of ambulation.
`Patients who are using a walker or wheelchair are scored
`similarly despite any other concomitant disability (eg,
`compromising cognitive dysfilnction). Regardless of the
`presence or absence of comorbid MS symptom maniw
`Fesrations such as pain. fatigue, vision abnormalities,
`sensory disturbances, or cognitive siowing, 2 patients
`each using a walker would have the same score.
`Secondly, at low numbers on the scale (below 3) there
`is significant
`inter- and intra-rater variability. Such
`variability has been posited to be, at least in part, relat-
`ed to Factors such as symptom fluctuations (widely rec-
`ognized as a common phenomenon in MS, especially
`with changes in body and ambient temperature, time
`of day, the season, following exertion, and with psy—
`chological stress;
`the so-called Uhthoff’s phenome—
`non) and the heterogeneity of assessment technique
`across different study examiners. Coupled with
`
`patient—reported subjective impressions of work per-
`formance, activities of daily living, and quaiity of life,
`such variability in the assessment of the neurologic
`examination over time powerfirlly underscores one of
`the most formidable challenges of ascertaining dynam-
`ic changes in disability. Unfortunately, when objec—
`tively trying to prognosticate for patients with MS, we
`are iimited by the few validated domains of efficacy
`data collected and analyzed From essentially all of the
`pivotal phase III clinical
`trials—principally reiapse
`rate, EDSS score, and radiographic measures of MS
`disease activity (magnetic resonance imaging (MRI]
`changes).
`
`CLINICAL Faerons Arrscrtuc B!SEASE RISK
`
`Multiple MS popuiation studies have analyzed
`patient demographic information relative to long—term
`disability in an attempt
`to identify "higli~risk”
`patients. These patients were more likely to have sig-
`nificant disability (quantified by the EDSS) over a 10—
`to 20-year period of time. Factors such as gender, age,
`ethnicity, and location of first attack have all been ana—
`lyzed to determine their relative prognostic signifi—
`cance.: Likewise,
`the assessment of “early" disease
`activity, as measured by relapse rate, has received sig—
`nificant attention for its utiiity to identify patients at
`higher (versus lesser) predilection for precocious dis—
`ability progressiou. Beyond demographic and clinical
`measures of disease activity, MRI metrics have been
`
`Table 1.. Demographics of Benign Multipie sclerosis
`
`
`
`Ramsaransing and De Keyser. 2007
`|
`Sayao et al. 2007 (20 years)
`Costelloe et al. 2008 (20 years)
`BenIgn
`Non-Benign
`P Value
`I
`Benign
`Non-BenIgn
`PValue
`Benign
`Non-BenIgn P Value
`
`(I 5 I)
`(345)
`|
`(88)
`(BI)
`(53)
`(as)
`|
`|
`| 275:8!
`
`3119.26
`
`.0l5
`
`28319.6
`
`34i§22
`
`.004
`
`
`
`Age
`
`30i8.8
`
`3Sil L6
`
`.0008
`
`96 female
`Pyramidal symptoms at onset
`Sensow change at onset
`ON at onset
`Brain stem at onset
`toss at 5 years
`EDSS at I0 years
`
`72. 1%
`23%
`42%
`37%
`I996
`13:09
`ND
`
`66.6%
`46%
`48%
`24%
`I6???»
`azure
`ND
`
`i
`l
`l
`
`>05
`.OOOI
`>05
`.003
`E
`>.05
`mom i
`|
`
`85.2%
`9.l%
`53. 96
`I9. 95
`5.9%
`ND
`
`72.8%
`8.6%
`50.6%
`I4. 95
`H.896
`ND
`
`.047
`
`£00053
`
`88%
`i
`96
`47%
`22%
`3 96
`
`i
`i
`E
`i
`g
`i
`i
`
`.006
`.OUI
`
`68%
`40%
`36%
`20%
`2 96
`
`EDSS = Expanded Usability Status Stale: ND = not done: ON = optic neunas.
`Data from Ramsaranssng and De Keyser‘; Sayao et a! : and Costelfoe et al."
`
`8
`
`Page 3 of 12
`
`Vol. 12, No. l. I March 2012
`
`Page 3 of 12
`
`

`

`PROCEEDINGS
`
`
`extensively invesrigated with respecr to similar prog—
`nostic capabilities. Ultimately, taken together, our clini-
`cal and tadiographic assessments potentially serve to
`stratify patients relative to their disease state. The coor-
`dinated strategy of analyzing multiparametric clinical
`and paraclinical outcome data modeling should be val»
`itiated with respect to being prognostically predictive
`of a lower versus a higher rislt for disease-related dis-
`ability. If confirmed, such models of assessment will
`porentiaily be integrated with novel information about
`individual patients (such as pharmacogenomic factors)
`that could be translated into corresponding and pre—
`cisely individualized and predictably effective treat—
`ment recommendations—a heretofore unprecedented
`advance in contemporary neurorherapeutics.
`Before, during, and following our meeting, several
`population studies were reviewed to determine what
`baseline clinical features were most associated with
`
`poor outcomes in MS. These included data from the
`Lyon MS cohort published by Confavreux et al
`in
`2003; a meta—analysis of studies published by Langer—
`Gould er al in 2006; and studies of“benign MS" pub-
`lished by Ramsaransing and De Keyser in 2007, Sayao
`et al in 2007, and Cosrelloe et al in 2008.” The major—
`ity of data sets suggest that male gender, older age,
`African American ethnicity, and moror symptoms at
`onset are associated with worse outcomes in MS (Table
`
`1). Furthermore, when early relapses were examined,
`more frequent
`relapses in the first 5 years,” and
`diminished time between events were associated with a
`
`higher likelihood of disability at epochs 10 and 15
`years after disease onset. Nevertheless,
`these studies
`consistently underesrimare the potential magnitude of
`accrued disability in MS because their outcomes and
`related conclusions are exclusively telescoped to a
`patient’s EDSS score. Conspicuously, between 19% to
`45% of patients designated as having so-called benign
`MS have been confirmed to exhibit evidence ofcogni—
`rive dysfunction. Difficulties with attention, word
`finding,
`information processing speed, multitasking,
`parallei processing, and executive planning comprise
`the broad diversity of inreilectual changes that can
`characterize “cognitive dysfunction" in MS, albeit
`despite being able to ambulate quite effectively and
`safely in many (ie, low EDSS). Thus, when consider-
`ing the true level of disability from MS, clinicians and
`patients should be aware of the constellation of poten—
`tial disease effects. Interestingly, in one study of cogni—
`tive impairment amongst “benign MS patients,” there
`
`
`were in faet concomitant correlations with higher lev-
`els of disease burden as measured by MRI.'
`
`MRI Facrons Arrecrmc DISEASE RISK
`
`Magnetic resonance imaging has become a corner-
`stone of clinical investigation assessment protocols of
`patients with MS. Since its first application to a
`patient with MS in 1981, MRI has literally revolu-
`tionized our ability to diagnose and monitor the MS
`disease process over time. Innumerabie technical and
`protocol
`refinements have markedly augmented the
`sensitivity and specificity of both brain and spinal cord
`lesional conspicuiry, which has thereby facilitated the
`capability of the neurologist to utilize highly precise and
`reproducible information about the dynamics and ulti-
`mate disposition (eg, the fate and destiny of a plaque
`lesion to proceed toward tissue destruction and the
`appearance and persistence of a black hole) of centrai
`nervous system tissue injury. Balancing the mechanisms
`of
`inflammation,
`demyeiination,
`remyelination,
`astrogliosis, axonal dysfunction (ion channel patho-
`physiology, perturbations in intermediate metabolic
`pathways in response to supply—demand mitochondrial
`energetic mismatch mechanisms, microrubuiar decon-
`struction, and neurofiiament disassembly, among
`other intra—axonal and intraneuronal derangements),
`axonal rransecrion, and neurodegeneration ultimately
`culminate in biasing the nervous system’s risk of per»
`manent injury versus the potential penchant for neu—
`roprotection, and even perhaps neuro—restoration. Not
`withstanding the impressive and pervasive progress
`achieved in the development of novel imaging para-
`digms, there has been a long-recognized clinical—radi—
`ographic paradox in MS that has yet
`to be fully
`explained. There are countless documented cases of
`patients with relatively minimal MRI—identified
`pathology, but significant disability, whereas airernare-
`ly there are patients with profound changes on MRI,
`albeit with relative preservation of neurologic func—
`tioning. Nonetheless, MRI has been shown to be
`informative about patient prognosis in several ways.
`Fisniltu et
`a] published outcomes data from
`approximately 80 patients followed for 20 years, and
`noted that increased lesion load at diagnosis was an
`independent prognostic indicator for precocious and
`more severe long-term disability as measured by
`EDSS.R What has been more controversiai is the pre-
`dictive value of asymptomatic white-matter lesions
`over time relative to disability. Data were reviewed
`
`
`
`Johns lloplrins Advance-ti Siutlies in b‘ietlicine I
`
`9
`
`Page 4 of 12
`
`Page 4 of 12
`
`

`

`PROCEEDINGS
`Mm—wuflu—mm—wwum—
`
`indicating that neariy 100% of T2 hyperintense
`lesions were gadolinium enhancing at some point,
`thus one would expect the prognostic significance of
`new T2 lesions to be similar to those that are enhanc»
`
`ing. Yet, when surveyed at the Philadelphia meeting,
`neurologists in the community were more concerned
`by the identification of gadolinium-enhancing
`lesions than with the presence of new T2 hyperin-
`tense lesions on routine surveillance MRls. A meta—
`
`analysis of individual patient data from 2 large,
`placebo-controlled clinical
`trials of subcutaneous
`interferon B-la in patients with RRMS or secondary
`progressive MS (SPMS) were anaiyzed separately and
`as pooled data to assess surrogacy for the number of
`new T2 hyperintense lesions. The number of new T2
`hyperintense lesions correlated with the number of
`relapses over the follow—up period. The proportion of
`treatment effect on relapses accounted for by the
`effect of treatment on new T2 MRI iesions over 2
`
`years was 53% in patients with RRMS and 67% in
`patients with SI’MS.q Another study tested the valid-
`ity of MRi surrogacy in MS studies on recently pub—
`lished trials of oral drugs!” Ninety«two percent of
`observed effects of oral drugs on clinicai outcomes
`could be predicted by the presence of active lesions
`on MRI.“ This further validates MRI surrogacy in
`MS, with important
`implications for
`individual
`patient management.
`In a meta‘analysis of 5 natural history studies and
`4 placebo—controlled clinical
`trials involving 307
`patients (RRMS = 237, SPMS a 70), Kappos et al
`found that neither gadolinium enhancement in the
`initial scan, not in 6 subsequent monthly scans, was
`predictive ofchange in EDSS score at 12 or 24 months
`(admittediy a relatively short epoch of time compared
`to the overall risk during the life of a patient with
`MS).“ The best predictor of relapse during the first
`and second years following diagnostic confirmation
`was change in the number of gadolinium—enhancing
`lesions on scans taken during the initial 6 months.
`Nevertheless, a recent meta—analysis study that includ-
`ed 23 randomized, double—blind, placebo—controlled
`trials in RRMS, for a total of 63 arms, 40 contrasts,
`
`and 6591 patients, showed that more than 80% of the
`variance in the effect on relapses between trials can be
`explained by the variance in MRI effects. Therefore,
`smaller and shorter phase ii studies based on MR}
`lesion end points may also give indications on the
`effecr of the treatment on relapse end points."
`
`OTHER CONSIDERATIONS 1N RISK AssessnrrNr
`
`Two important characteristics about patients were
`repeatedly highlighted in the patient management ses—
`sions at the Phiiadelphia meeting. While duration of
`disease figured prominently in rendering disease—
`modifying treatment decisions for patients with MS,
`relapses or MRI changes occurring early in the course
`of the disease were considered more significant from a
`disease risk perspective than insidious changes occur-
`ring years into the course of the disease. This duration
`ofdisease—based impact upon prognosis has previously
`been quantified by the MS Severity Score, published
`in 2005 by Roxburgh er al.‘2 Another important and
`influential
`factor with respect
`to treatment was a
`patient’s baseline defining characteriStics with respecr
`to the ievel of clinical disability, and the MRI burden
`of disease. Those with pre—exisring disability,
`large
`MRI burdens of disease, or lesions located in eloquent
`regions that represent harbingers for more substantial
`disability (cg, brain stem and spinal cord) were strati—
`fied into higher risk designations that justified a com-
`mensurate intensification of immune modulatory
`therapy, when compared to those classified into lower
`risk categories.
`After reviewing the available data, neurologist par-
`ticipants in the MS Think Tank meeting applied the
`derived principles to teak-world patient scenarios.
`Patients were risk stratified relative to their disease
`
`characteristics, followed by specific treatment recom—
`mendations for each respective patient. Although there
`was not a singular consensus approach in the manage—
`ment for each scenario (nor was this the objective of
`the meeting), severai important and highly salient
`themes were codified within a treatment framework
`
`that will be underscored throughout the analysis of
`each patient vignette considered within this article.
`The proposed management for each patient with MS
`is based on identifying a disease-modifying therapy
`that would be anticipated to adequately suppress dis-
`ease activity within a given patient, and based on dif-
`ferential and individualized considerations germane
`to balancing both efficacy and rislt of the selected
`treatment.
`
`RISK ASSESSMENT or
`FDA-APPROVED Tusmrtes IN MS
`
`There are currently 8 FDA-approved therapies for
`RRMS (Table 2). Their approval was based on pivotal
`randomized placebo—controlled phase III
`trials that
`
`
`
`10
`
`Page 5 of 12
`
`lol. 12, No.
`
`1 a March 20 2.2
`
`Page 5 of 12
`
`

`

`PROCEEDINGS
`
`
`Tabie 2. FDA-Approved Therapies for Reiapsing—Remitting Multiple Scierosis
`
`Interferon
`lnterferon
`
`B-i b”
`[3- ia
`
`Giatimmer
`
`Acetate
`
`Mitoxantrone
`[5- la
`
`Interferon
`
`Nataiizumab
`
`Fingolimod
`
`370A approval
`Route
`
`Frequency
`Reduction in relapses
`Placebo ARR
`
`1993
`SC
`
`Q05
`3 E96
`£27
`
`i996
`1M
`
`Weekly
`32%
`0.9
`
`|996
`SC
`
`Daily
`29%
`0.84
`
`2000
`
`IV
`
`Quarteriy
`65%
`i.3i
`
`2002
`
`SC
`
`TIW
`32%
`LZB
`
`2004/2006
`
`lV
`
`Monthiy
`66%
`0.73
`
`20 | 0
`
`PO
`
`Daily
`60%
`0.4
`
`NNT to prevent a relapse
`3.2 3.4 3.l
`(relative to ARR)
`
`
`
`
`
`
`
`3. i l .5i .5 | .7
`
`
`
`
`
`‘There are 2 manufacturers ofthis cling: the only dIrTerence IS the brand name of the drugs.
`ARR : annualized relapse rate; FDA = US iood and Drug Administration: "‘1 = intramuscular: iv = intravenous; NNT = number needed to treat PO 3 by mourn. Q01) 3
`every other clay. SC : subcutaneous, TlW : 3 umes weekly.
`
`captured data relative to relapse rate suppression, dis-
`ability progression, and MR1 outcomes. In all cases,
`these therapies outperformed the placebo arm, leading
`to their FDA approval and widespread utilization.”
`Despite the availability and ever—increasing use of
`these agents (including an expanding number of early
`adopters to treatment at the time of a first ciinically
`isolated inflammatory demyeiinating syndrome, as
`weli as increased application of these treatments in
`those with radiographic signatures of MS, ptcempting
`even the first clinical event, so«calied radioiogicaily iso-
`lated syndrome),
`the ultimate impact of disease-
`modiFying therapy on the disposition of productive
`life and quality of life remains enigmatic.
`Fir5t, there is no consensus on what constitutes the
`
`“ultimate outcome" against which “success" should be
`measured.
`Ideally, a therapeutic intervention would
`prevent patients from having any new motor, sensory,
`bowel/bladder, visual, pain,
`fatigue, or cognitive
`changes from MS. To prove this over the long term,
`large numbers of patients wouid have to be systemati—
`cally foliowed for decades, something that is patently
`untenable within the context of a ciinicai trial, and like-
`
`ly impossibie outside of the consrrucrs and ascertain—
`ment architecture of clinical
`investigation.
`Instead,
`short-term outcomes (relapse rate, short—term disability
`progression, and MRI measures) are analyzed within the
`rigors ofcontrolied trials, and are thereby posited to rep—
`resent surrogates of the more important, bona tide, and
`ultimately informative,
`longer phases oi: the disease
`course and its patient-specific trajectory. Second, how
`
`MS impacts upon surrogate measures of disease activity,
`and how this impact should be expressed, has not yet
`been universally codified. For example, the reduction in
`relapse rate experienced by patients randomized to the
`treated arm within a trial can be expressed in several
`ways including absolute risk reduction,
`relative risk
`reduction, percent of patients who are rciapse free, or
`the number needed to treat (NNT) in order to prevent
`one relapse. Each of these measures of relapse rate
`change provides a somewhat different outcome metric
`with respect to a drug’s efficacy.
`Idealiy, csrablishing methods in order to elucidate
`patterns of drug effect across clinical
`trials would
`greatly Faciiitate both patients and physicians in opti—
`mizing the process of treatment selection. The MS lit-
`erature is
`replete with more than 2 decades of
`randomized controlled clinicai triais that could form
`
`the basis of a rigorous interrogation of the evidence in
`order to ascertain the differential magnitude of effica—
`cy to be derived From each of the FDA-approved
`agents. Yet, the relative paucity ofhead—to—hcad drug tri-
`als, and the changing phenotype ofpiaccbo arm patients
`over time, makes direct comparisons between placebo-
`controlied trials untenable. Differences in the event rate
`
`among placebo arms can have profound impacts on
`outcome measures such as relative risk and NNT.
`
`For example, in 2 trials where treated patients had
`an absolute risk reduction of 0.4 relapses per year corn—
`parcd to placebo, the relative risk reduction changes
`based on the acrual event rates. In one trial, For exam-
`
`ple, where the treatment arm had an ARR of 0.5 and
`
`
`
`Johns linplrins Advanced Studies in Medicine I
`
`Page 6 of 12
`
`11
`
`Page 6 of 12
`
`

`

`i’ROCEEDINGS
`
`
`the placebo arm an ARR of 0.9 (absolute risk reduo
`tion a 0.9 ~ 0.5, or 0.4), the reduction in relative risk
`
`of relapse would be 1 - (0.5/0.9), or 44. %. In a sec-
`ond trial, with the same absolute risk reducrion of 0.4
`
`relapses per year, an ARR of0.1 in the treated arm and
`0.5 in the placebo arm, the reduction in reiative risk is
`an impressive 80%. Thus, the relative activity of the
`enrolled patient population can have a profound
`impact on the conclusions drawn when using relative
`risk reduction analysis. This provides an example of
`the challenges we face when trying to compare relative
`efficacy of different therapeutic agents.
`The effects of a changing placebo arm phenotype can
`impact the NNT, but only at the extremes of event rates.
`For example, if a trial has a placebo arm event rate of
`85% and a treated arm event rate of 70%, the absolute
`risk reducrion would be 15% and the NNT would be
`
`1/015, or 6.67. Thus, you would need to treat between
`6 and 7 patients for one year to prevent one event.
`However, a trial with a placebo arm event rate of 10%
`and a treatment arm event rate ofO (in essence a cure rel-
`ative to that event), would have an NNT of 10 (1/01).
`Thus, while the second drug would appear to have a
`iower absolute risk reduction of 10% and a higher NNT,
`it was in essence a cure relative to the low event rate of
`
`the population enroiled in the placebo arm. The risk of
`inappropriate conclusions using NNT is highest when
`placebo arm rates are relativeiy high. The NNT data and
`placebo event rates for currently available disease-modi—
`fying therapies are iisred in Table 2.
`
`Thus, there is no easy way to compare data about
`agents generated through dilierent trials. The lack of
`well-designed head-to-head trials limits our ability to
`accurately rank disease~modjfying therapies from an efli—
`cacy perspective. Based on available data and experience,
`participants at the meeting were able to group medica—
`tions into categories based on NNT trends. This
`approach focused on the percent change in disease activ-
`ity that initiation of a drug would lead to. Beyond eli—
`cacy, however, a comparison of safety is easier from
`available data. While unrecognized biases invoived in
`study enrollment could alter adverse event rates,
`this
`seerns less iii-rely than alteration of clinical event rates.
`An anaiogous measure to NNT relative to adverse
`events is the number needed to harm (NNH). This fig—
`ure reports the number of patients needed to treat before
`an adverse event occurs. in an ideal world, a drug would
`have an NNT of 1 and an infinitely high NNI—i, repre-
`senting a cure without risk. Similar to NNT, the NNH
`is calculated by dividing 1 by the absolute risk difference
`of: an adverse event. For example, if 60% of patients on
`interferon had flu-like symptoms, compared to 20% of
`patients on placebo,
`the absolute risk attributable to
`interferon would be 40%. Thus, the NNH would be 2.5
`
`Interpreting the clinical significance of the
`(1/04).
`NNl-I is always relative to the severity of the adverse
`event being considered.
`in
`During the MS Think Tank meeting
`Philadelphia, relative rates of serious adverse events For
`each of the FDA-approved diseaseamodifying therapies
`
`Table-i3. NNH foe-Serious Adverse Events ititiFDA—Approved Disease-ModifyingTherapies '-
`
`l Mitoxantrone
`Natalizumab
`Fingolimoci
`Giatiramer Acetate
`Interferon IM
`Interferon SC
`
`N/A
`l
`MIA
`]
`NIA
`IBZ
`N/A
`25—33
`Injection Site necrosas
`NJ'A
`N/A
`i
`N/A
`N/A
`500
`500
`Suicide attempt
`N/A
`N/A
`N/A
`8
`MA
`MA
`Pear-injection flushing
`Reduction in ejection fraction
`NJ'A
`NJ'A
`NI'A
`8
`N/A
`N/A
`
`Leukemia
`N/‘A
`NI'A
`N/A
`36—400
`N/A
`NIA
`
`AV conduction block
`Macuiar edema
`
`N/A
`NA
`
`IOOO
`250
`
`
`
`
`NJ'A
`N/A
`
`l
`
`N/A
`N/A
`
`N/A
`NJ'A
`
`NJ'A
`N/A
`
`N/A
`25
`NA
`NA
`N/A
`N/A
`Infusion reaction
`N/A
`soo"
`N/A
`N/A
`N/A
`1
`N/A
`PML
`P
`P
`Y
`P
`8
`E
`.
`ataents ex osed to natailzumab. lndmduai nsk ma be in her or iower de endln on the duration of natahmrnab use. history of pnor immunosuo-
`"Toss estimate is for all
`pressron use. and anu-jakob-Creutzfeidt 0C) vu‘us antibody status.
`AV 2 amovenmcular. IN = intramuscuiar. MIA 2 not applicable: NNH = number needed to harm: PML = progresswe muiufotal ieukoencephaiopathy: SC 2 subcutaneous.
`
`l
`
`
`
`Vol. i2, ND.
`
`1 I March 2012
`
`Page 7 of 12
`
`Page 7 of 12
`
`

`

`I’ROCEEDiN‘GS
`
`
`were reviewed. Some of the NNH statisrics are based
`
`on variable estimates in the literature and are repre-
`sented in Table 3 as a range. These data are
`primarily derived from event rates that are summarized
`in
`the medications
`prescribing
`information
`Determining which patients are at highest risk for var-
`ious serious adverse events would be quite useful in
`deciding when to use various agents. \Vith the advent
`of anti—Jakob—Creutzfeldt (IC) virus antibody testing,
`patients and physicians have