`
`Intermittent cyclophosphamide
`pulse therapy in progressive
`multiple sclerosis:
`Final report of the Northeast Cooperative
`Multiple Sclerosis Treatment Group
`
`H.L. Weiner, MD; G.A. Mackin, MD; E.J. Orav, PhD; D.A. Hafler, MD; D.M. Dawson, MD;
`Y. LaPierre, MD; R. Herndon, MD; J.R. Lehrich, MD; 8.L. Hauser, MD; A. Turel, MD; M. Fisher, MD;
`G. Birnbaum, MD; J. McArthur, MD; R. Butler, MD; M. Moore, MD; B. Sigsbee, MD; A. Safran, MD;
`and the Northeast Cooperative Multiple Sclerosis Treatment Group*
`
`
`
`Article abstract—Previous studies reported that a 2- to 3-week course of IV cyclophosphamideplus adrenocorticotropic
`hormone (ACTH) induction can temporarily halt progressive MS for a period of 12 months in the majority of patients
`treated, after which reprogression occurs. The Northeast Cooperative Multiple Sclerosis Treatment Group was formed
`to determine whetheroutpatient pulse cyclophosphamide therapy could affect reprogression and whetherthere weredif-
`ferences between a modified induction regimen and the previously published regimen. Two hundredfifty-six progressive
`MSpatients were randomized into four groups to receive IV cyclophosphamide/ACTH via the previously published ver-
`sus a modified induction regimen, with or without outpatient IV cyclophosphamide boosters (700 mg/m? every other
`month for 2 years). There were blinded evaluations performed every 6 months. Results demonstrate that (1) there were
`no differences between the modified and the published induction regimenseither in termsofinitial stabilization or sub-
`sequent progression; (2) without boosters, the majority of patients continued to progress; and (3) in patients receiving
`boosters, there was a statistically significant benefit at 24 months and 30 months(p = 0.04). Time to treatment failure
`after 1 year wasalso significantly prolonged in the booster versus the nonbooster group (p = 0.03). Age was the most
`important variable that correlated with response to therapy in that amelioration of disease progression occurred primar-
`ily in patients 40 years of age or younger. Boosters had a significant benefit on time to treatmentfailure in patients ages
`18 to 40, p = 0.003, but not in patients ages 41 to 55, p = 0.97. In addition, patients with primary progressive MS had a
`poorer prognosis at 12 months than patients with secondarily progressive MS(p = 0.04). Ourfindings (1) support a role
`for immunosuppression in the treatment of MS, (2) begin to identify variables that may explain differences between
`studies of immunosuppression with cyclophosphamidein progressive MS, and (3) suggest that intermittent pulse thera-
`py is an important method for the treatment of progressive MS and perhapsfor earlier stages of MSas well.
`NEUROLOGY1993;43:910-918
`
`Multiple sclerosis is an inflammatory disease of the
`central nervous system of presumed autoimmune
`etiology. There are a number of immune abnormal-
`ities in MS, including loss of suppressor influences
`and activated T and B cells both in the CNS and
`the peripheral blood.! The design of the majority of
`immunotherapeutic approaches studied over the
`past 20 years has been to suppress the immune
`
`system in patients with MS with both antigen-spe-
`cific and antigen-nonspecific suppression.?
`Based on uncontrolled reports which suggested
`that intensive immunosuppression with short-term
`administration of cyclophosphamide plus cortico-
`steroids affected the course of progressive and
`relapsing-remitting MS,?’ in 1980, Boston investi-
`gators undertook a randomizedtrial of high-dose
`
`
`
`* For list of centers, see Appendix on page 918.
`From the Multiple Sclerosis Unit of the Center for Neurologic Diseases, Division of Neurology, Department of Medicine (Drs. Weiner, Hafler, Mackin,
`and Dawson), Brigham and Women’s Hospital, Harvard Medical School, Boston, MA; Department of Biostatistics, Harvard School of Public Health (Dr.
`Orav), Boston, MA; Montreal Neurologic Institute (Dr. LaPierre), Montreal, PQ, Canada; Massachusetts General Hospital (Drs. Lehrich and Hauser),
`Boston, MA; Strong Memorial Hospital (Dr. Herndon), Rochester, NY; Geisinger Medical Center (Dr. Turel), Danville, PA; University of Minnesota
`Hospital (Dr. Birnbaum), Minneapolis, MN; Worcester Memorial Hospital (Dr. Fisher), Worcester, MA; Johns Hopkins Medical Center (Dr. McArthur),
`Baltimore, MD; Eastern Maine Medical Center (Dr. Sigsbee), Bangor, ME; Framingham Union Hospital (Dr. Safran), Framingham, MA; and Emerson
`Hospital and Boston University School of Medicine (Drs. Butler and Moore), Concord, MA.
`Supported by a grant from the National Multiple Sclerosis Society.
`Presented in part at the 43rd (Boston, MA} and 44th (San Diego, CA) annual meetings of the American Academy of Neurology, April 1991 and May 1992.
`Received March 16, 1992. Accepted for publication in final form September 14, 1992.
`Address correspondence and reprint requests to Dr. Howard L. Weiner, Center for Neurologic Diseases, Longwood Medical Research Center, 221
`Longwood Avenue, Boston, MA 02115-5817.
`.
`
`910 NEUROLOGY43 May 1993
`
`1
`1
`
`Hopewell EX1021
`Hopewell EX1021
`
`

`

`intravenous cyclophosphamide plus adrenocorti-
`cotropic hormone (ACTH) versus ACTH alone or a
`plasma exchange regimen in 58 patients with pro-
`gressive MS. Results demonstrated that a 2- to 3-
`week course of IV cyclophosphamide plus ACTH
`could halt the progression of chronic progressive
`MSfor a 1-year period in approximately 75% of the
`20 patients treated.’ Continued follow-up demon-
`strated that of those patients who benefitted from
`treatment either by stabilization or improvement,
`the majority began reprogressing between 12 and
`18 months after the initial treatment.? However,
`there were patients whose reprogression began 6 to
`9 months after treatment or as long as 3 years
`after treatment. This study supported the positive
`results obtained by earlier investigators who had
`treated MS patients with cyclophosphamide in
`combination with corticosteroids.?7
`Following publication of the Boston results in
`1983, there was heightened interest in the poten-
`tial use of immunosuppression with cyclophos-
`phamide in progressive MS. The Northeast
`Cooperative Multiple Sclerosis Treatment Group,
`formed in 1984 to further study the effect of
`cyclophosphamide on the course of progressive MS,
`allowed multiple centers to treat patients according
`to a standardized protocol and to treat sufficient
`numbers of patients to answertheclinical ques-
`tions raised in the protocol. The Treatment Group
`had several options for the next questions to inves-
`tigate regarding the use of cyclophosphamide in
`progressive MS, including whether concomitant
`steroids were required and whethertheoriginal
`study should be repeated. The group decided that
`the most important question was whether a form of
`intermittent outpatient pulse cyclophosphamide
`therapy, that has become standard for diseases
`such as lupus nephritis,’ could affect subsequent
`reprogression in progressive MS patients treated
`with cyclophosphamide/ACTH induction. In addi-
`tion, we wished to investigate whether a modified
`regimen that required shorter hospitalization, or
`that could be administered on an outpatient basis,
`wasasefficacious as the published regimen.
`
`Methods. Patient population. A total of 261 eligible
`patients with progressive MS were randomized among
`four treatment groups. This numberexcludes 26 patients
`with no data who were randomized but not treated
`because of subsequent issues concerning eligibility. An
`additional five patients were removed from all analyses
`because no data from the initial examination were submit-
`ted, making it impossible to assess progress (two of the
`five patients were from arm 1, and one patient was from
`each of the other arms). The remaining 256 patients pro-
`vided data for analysis. The original study protocol called
`for 75 patients per arm in order to detect 15% stabilization
`or improvement on the nonbooster arms at 3 years com-
`pared with 40 to 45% stabilization or improvement on the
`booster arms, with 95% power and 5% type I error.
`Patients had clinically definite MS according to the
`Schumachercriteria with at least 1-point worsening on
`the Kurtzke Disability Status Scale (DSS) or Ambulation
`
`Index (AI) in the 12 monthsprior to entry. Screening by
`history and laboratory tests where appropriate was car-
`ried out before entry to rule out other diagnoses such as
`systemic lupus erythematosus and Sjégren’s syndrome.
`Worsening prior to entry was determined by examination
`of patients (change in AI) or history and record review if
`changes involved clear worsening on DSS (eg, need for
`the use of a cane). At entry, patients were classified as
`having primary progressive MS (progressive MS from
`onset of disease without a history of relapses or remis-
`sions) or progressive MS that evolved after a prior
`relapsing-remitting course. Patients were between the
`ages of 18 and 55 at randomization and had a DSSof 3
`through 6B (6B = Expanded Disability Status Scale
`[EDSS] of 6.5, requiring bilateral support for ambula-
`tion) or a DSS of 7 that occurred due to disease progres-
`sion in the previous 2 months.
`The study was designed to include participation of both
`academic institutions and private neurologic practices
`since the treatment involved readily available drugs. Prior
`to initiating the trial and at periodic times during the
`course of the study, investigator meetings were held to
`assure standardization of neurologic assessments and con-
`duct of the trial. A total of 28 participating groups started
`the trial, although nine centers were removed when the
`majority of their data remained incomplete. The remain-
`ing 21 centers enrolled patients that were randomized cen-
`trally at the Brigham and Women’s Hospital in Boston.
`Review of initial evaluations, verification of eligibility,
`checks for consistency of data, and maintenance of the
`database were performed centrally by neurologists (G.A.M.
`and H.L.W.) at the Brigham and Women’s Hospital.
`Treatment regimens. There were four treatment groups.
`All patients received induction with cyclophos-
`phamide/ACTHin the hospital. Groups 1 and 2 were ran-
`domized to receive treatment according to the previously
`published regimen: 125 mg cyclophosphamideintravenous-
`ly four times a day over 8 to 18 days until the white blood
`cell count fell below 4,000/mm? plus IV ACTH.§ Groups 3
`and 4 received a modified regimen in which cyclophos-
`phamide at a dose of 600 mg/m? was given intravenously
`on days1, 2, 4, 6, and 8. Patients on the modified regimen
`received IM ACTH over 14 days (40 units twice daily for 7
`days, 40 units daily for 4 days, and 20 units daily for 3
`days). Following induction in the hospital, patients in
`groups 2 and 4 received 700 mg/m? cyclophosphamide
`intravenously every 2 monthsfor a 2-year period, whereas
`patients in groups 1 and 3 remained untreated.
`Neurologic evaluation and conduct of the study. The
`protocol was approved by institutional review boards at
`each of the participating centers. After informed consent
`was obtained, patients were randomized centrally at the
`Brigham and Women’s Hospital and hospitalized at their
`respective center. Individual randomization schemes were
`prepared for each center and kept at the Brigham and
`Women’s Hospital, and assignment to an experimental
`group occurred at the time of randomization. Specifically,
`when a center identified a patient eligible for the study,
`they contacted the Brigham and Women’s Hospital and
`were informed of the assigned treatment group. Twoscor-
`ing systems were used to assess neurologic status: (1) DSS
`in which the sixth category was divided into 6A (unilateral
`support for ambulation) or 6B (bilateral support for ambu-
`lation)—these categories are equivalent to 6.0 and 6.5 on
`the EDSS; and (2) the AI.® Patients were evaluated on
`admission to the hospital and at 6-month intervals for 3
`years thereafter. In addition, an evaluation was performed
`whenever a patient reported worsening of the condition.
`
`May 1993 NEUROLOGY43 911
`
`2
`
`

`

` Table 1. Patient demographic andclinical
`
`No
`boosters
`
`129
`39.84 8.1
`60%
`
`Boosters Modified
`
`Published
`
`127
`40.3484
`59%
`
`139
`40.448.1
`60%
`
`36
`27
`88%
`
`36
`29
`78%
`
`34%
`40
`26
`81%
`
`81.7475
`
`31349.4
`
`31.5485
`
`31.64 8.5
`
`9%
`6
`44
`32
`10
`59409
`
`14%
`49
`33
`4
`
`é
`
`e
`
`¢
`
`Q
`
`14%
`4
`40
`32
`10
`6.8411
`
`18%
`45
`33
`3
`
`a%
`10
`37
`37
`8
`5.9408
`
`16%
`44
`38
`3
`
`characteristics
`
`Sample size
`Average age (+SD)
`Sex: % women
`
`Type of MSpriorto
`onsetof progression
`Relapse/remit
`Relapse/prog.
`Chronic onset
`Prior treatment with
`ACTH, prednisone
`Average age at onset
`Disability score
`3-4
`5
`6A
`6B
`7
`Average
`Ambulation Index
`
`Average
`
`Neurologic evaluations were performed in a single-blind
`fashion by the examining neurologist. For most centers,
`except the Brigham and Women’s Hospital, there was a
`single examining neurologist for each patient and this
`individual was also the treating physician. All patients
`experienced alopecia, which was evident at the 6-month
`evaluation. By 12 months there was no alopecia, and the
`dose of cyclophosphamide boosters did not result in hair
`loss; thus, a blinded examination could be carried out dur-
`ing the entire study. Although formal interrater variabili-
`ty was not assessed, all examining neurologists were
`familiar with and had experience using the scales, and
`detailed discussion of rating methods were carried out
`prior to and during the course of the study.
`Statistical methods. This study was designed to
`answer two questions: (1) How does the administration of
`cyclophosphamide boosters in addition to induction
`(groups 2 and 4) compare with induction alone (groups 1
`and 3), and (2) how does a modified cyclophosphamide
`regimen (groups 3 and 4) compare to the published regi-
`men (groups 1 and 2)? To address these questions, three
`primary endpoints were chosen. Each of the endpoints
`was based on the following protocol-based definition of
`treatment failure: patients were treatment failures if
`they declined 1 point on the DSS and remained at that
`level for 2 months, were removed from follow-up for med-
`ical reasons related to treatment, or were removed from
`follow-up because of deviation from the protocol-pre-
`scribed treatment regimen. These last two components of
`failure were included so that we could analyze by intent
`to treat. A 1-point decline in the DSS included a change
`from 6A to 6B and from 6B to 7. In 18 instances (seven
`booster patients and six nonbooster patients), a patient
`declined 1 point, was not retreated, and recovered later
`in the study to be stable or improved. These patients
`were classified as treatment failures at the time of their
`decline for purposes of survival analysis. They were,
`however, included in calculations of percentage of
`patients improved or stable. All patients who were
`retreated either with steroids or immunosuppressive
`therapy remainedfailures for the rest of the study.
`To assess early differences, the first endpoint com-
`pares patient groups at 12 months on the basis of failure
`versus stabilization or improvement. Those patients who
`were not treatment failures at 12 monthswereclassified
`either as stable if their DSS was the sameasat the start
`of treatment, or as improved if their DSS improved by 1
`or more points. The Mantel-Haenszel trend test was used
`to compare patients on published induction to those on
`modified induction and to compare patients on boosters
`to those not on boosters.
`The second endpoint assessed long-term efficacy and
`is identical in definition and in terms of analysis to the
`first, except that 24-month data are used to define fail-
`ure, stabilization, and improvement. Timetofirst failure
`was the third endpoint, and allows analysis across the
`entire 3 years of follow-up. A proportional hazards sur-
`vival model was used to compare patient groups, both
`before and after adjustment for baseline characteristics.
`Patients who withdrew from the study for nonmedical
`reasons are treated as censored in such an analysis, con-
`tributing information only until their time of withdrawal.
`All analyses were repeated using 1-point changes on
`the AI to define failure, and for the first two endpoints,
`stability and improvement. Since these analyses pro-
`duced very similar results to those based on DSS, we
`refer only occasionally to the comparative results. In
`addition to the three primary endpoints, we compared
`
`912 NEUROLOGY43 May 1993
`
`3
`
`patient groups at each 6-month follow-up and used an
`ordinal logistic regression model to identify subgroups of
`patients moreproneto stabilization and improvement.
`Whenweidentified a characteristic, such as age, that
`was related to booster efficacy, we divided the patients
`by median age into two subgroups and analyzed each
`subgroup separately for the effect of boosters using the
`Mantel-Haenszel trend test and the survival regression.
`
`Results. Patient characteristics. The four treat-
`ment groups were well matched with respect to
`age, sex, and duration of disease, type of MS before
`onset of progression, previous treatment with
`ACTHor prednisone, and initial DSS and AI. The
`majority of patients, 73%, required unilateral or
`bilateral support for ambulation (DSS of 6A or 6B).
`No significant differences were found on any of
`these measures(table 1). The four groups werefol-
`lowed to failure or censoring an average of 508
`days, 510 days, 506 days, and 555 days. Patient
`accrual by center is shown in the Appendix.
`During the course of the 3-year trial, 26 patients
`withdrew due to medical complications from treat-
`ment or due to problems with treatment. Those
`patients were treated asfailuresas of their dates of
`withdrawal and throughout the remainderof the
`study. Six of these failures were among patients
`not on boosters (two on arm 1 and four on arm 3),
`while 20 of the failures were among boosters
`patients (10 on arm 2 and 10 on arm 4). An addi-
`tional 11 patients withdrew for reasons determined
`to be unrelated to treatment or disease, but all
`these patients had suffered disease progression and
`were therefore classified as failures for purposes of
`
`

`

`Table 2. Percentages of patients who were stable/improved measured by the Kurtzke Disability Status
`Scale comparing arms receiving maintenance boosters and no maintenance boosters
`
`Noboosters
`(groups 1 and 3)
`Maintenance boosters
`(groups 2 and 4)
`
`Significance level for
`a trend test comparing
`the two groups on the
`proportionsof patients
`improved,stable,
`and worse
`
`Ambulation Index showssimilar significant differences at 24 months (p = 0.04) and at 30 months(p = 0.03).
`
`
`
`
`SurvivalProbability
`
`survival analysis. Finally, 15 patients withdrew for
`nonmedical reasons during the first 6 months, and
`12 or 13 during each subsequent 6-month follow-up
`without evidence of disease worsening. These
`patients stopped contributing information to the
`analysis as of the dates they withdrew. They were
`evenly distributed between the booster (n = 34)
`and nonbooster groups (n = 29).
`Comparison of published versus modified induc-
`tion. No differences in response were found
`between patients receiving the modified induction
`compared with those receiving the published induc-
`tion. At 6 months, 22% were improved and 49%
`stable on modified induction compared with 24%
`improved and 52% stable on the published induc-
`tion (p = 0.43). Twelve-month data showed 19%
`improved on the modified arm and 35% stable, ver-
`sus 23% improved and 35% stable on the published
`arm (p = 0.48). At 2 years, 13% of modified-arm
`patients were improved and 17% werestable, ver-
`sus 9% improved and 21% stable on the standard
`arm (p = 0.91). The time-to-failure analysis showed
`similar results with virtually coincidental survival
`curves and a nonsignificant difference (p = 0.83)
`between the two induction regimens. Similar
`results were obtained on calculations using the AI.
`Comparison of maintenance boosters versus no
`boosters (table 2, figure 1). Maintenanceboosterssig-
`nificantly slowed progression at 24 months (p =
`0.04), although no improvement was noted at 12
`months (p = 0.71) or through the use of survival
`analysis (p = 0.18). At 24 months, the booster arm
`showed 16% improved and 22% stable, compared
`with 9% improved and 15% stable on the nonbooster
`arm. Further analyses verified that the impact of
`boosters persisted at 30 months (p = 0.04) and could
`be replicated on the AI at both 24 months (p = 0.04)
`and 30 months (p = 0.03). Table 2 suggests that the
`impact of booster therapy does not begin until after
`the 18-month evaluation. We pursued this idea by
`repeating the survival analysis, treating booster
`therapy as a time-varying predictor whose impact
`begins at 1 year. Figure 1 shows the comparison of
`
`TIME TO TREATMENT FAILURE
`
`All Patients
`
`Boosters
`
`Figure 1. Kaplan-Meier survival curves comparing time
`to treatment failure in patients receiving bimonthly
`cyclophosphamide boosters with patients receiving no
`booster therapy. Percentage of individuals who were not
`treatment failures are plotted versus time. No significant
`difference was found (p = 0.18) over the entire course of
`follow-up, but in examining booster effects starting at 1
`year, a significant benefit (p = 0.03) was detected.
`
`time to failure with and without boosters. Boosters
`had significant impact (p = 0.03) on delaying repro-
`gression after 1 year; this is consistent with previous
`observations that the average time to reprogression
`following cyclophosphamide/ACTH induction is
`approximately 18 months.° A similar survival analy-
`sis supported the finding using the AI to definefail-
`ure (p = 0.06). The actual values of the DSS at the
`follow-up times are not given as they are misleading,
`since patients who fail either have no DSS value or
`have a value that reflects retreatment as well as the
`original randomized therapy. For this reason, we do
`not report or compare these values.
`Comparison of different centers and identifica-
`tion of responsive subgroups. A large population of
`patients were treated at the Brigham and Women’s
`
`May 1993 NEUROLOGY43 913
`
`4
`
`

`

`Table 3. Effect of treatment in younger (ages 18-40) versus older (ages 41-55) patients*
`
`
`
`6 mo
`
`12 mo
`
`Percentage of patients improved or stable
`18 mo
`24 mo
`30 mo
`
`36 mo
`
`Patients on boosters (n = 107)
`
`Young (n = 54)
`Improved
`Stable
`
`Old (n = 53)
`Improved
`Stable
`
`p value
`
`Patients not on boosters (n = 113)
`
`Young (n = 53)
`Improved
`Stable
`
`Old (n = 60)
`Improved
`Stable
`
`p value
`
`81%
`26
`55
`
`10%
`18
`52
`
`0.14
`
`27
`27
`
`38%
`8
`30
`
`0.02
`
`12
`11
`
`25%
`6
`19
`
`0.79
`
`
`
`* Improvedor stable defined by DSS; at 36 months, analysis involves 81 patients on boosters (40 young, 41 old) and 94 patients not on boosters (47
`young, 47 old).
`
`Hospital (85 of 256); some centers treated very
`small numbers of patients (eg, two). Thus, data
`were analyzed according to center size to determine
`whether treatment was differentially successful at
`the Brigham and Women’s Hospital and at other
`centers. Since the primary finding of the study
`group was a slowing of progression at 24 and 30
`months in patients receiving boosters, these two
`time points were the focus of analysis. At 24 and 30
`months, a positive effect of boosters was observed
`both in the 85 patients treated at the Brigham and
`Women’s Hospital and in the 145 patients treated
`at centers with only 8 to 21 patients (p = 0.02 at 24
`months and p = 0.04 at 30 months). No effects of
`boosters were observed when centers with six or
`fewer patients were analyzed (n = 26).
`Wethen performed analyses to determine
`whether there were any other characteristics that
`were prognostic of success and whetherthe effects
`of boosters were specific to a particular subgroup of
`patients. Because of our previous findings, the
`impact of boosters on the survival analysis was
`allowed to appearonly after the first year of follow-
`up. We considered the following measures of clini-
`cal status at the beginning of treatment: DSS, AI,
`and each of the six functional status scales that
`focus on a specific locus of disability. We found that
`none of these predicted failure at 12 months, 24
`months, or throughout the course of follow-up.
`However, patients who were chronic progressive
`from the onset of their MS had particularly poor
`prognosis at 12 months (p = 0.04). Over 55% of
`patients who were chronic progressive from onset
`had failed by 12 months, while only 41% of the
`
`914 NEUROLOGY43 May 1993
`
`other patients had failed. The type of MS at onset
`was not prognostic at 24 monthsor in the survival
`analyses. Similarly, previous treatment with
`ACTHorsteroids did not alter prognosis.
`The most striking finding, however, was that
`younger patients early in their disease were most
`likely to stabilize or improve. Since the median age
`of patients in the study was 41 years, the clinical
`course of patients above and below the median was
`analyzed. In addition to age, a numberof other
`measures of “early in disease” were analyzed,
`including patients who had anearly onset (chronic
`progression before age 32), patients who had a
`recent onset of chronic progression (within the past
`7 years), and patients who were young and had
`recent onset (age less than 41 and chronic progres-
`sion within the last 7 years). Table 3 shows analy-
`ses of patients ages 18 to 40 versus patients ages
`41 to 55. Boosters are of greater benefit in younger
`patients (40% stable or improved at 30 months
`with boosters versus 9% without boosters, p =
`0.01) than in older patients (14% stable or
`improved at 30 months with boosters versus 25%
`without boosters, p = 0.27). Furthermore, at 18
`months, the percentage of patients improvedor sta-
`ble that did not receive boosters was also greater in
`younger than older patients (54% versus 38%, p =
`0.02). Figure 2 shows the comparison of time to
`failure with and without boosters among patients
`ages 18 to 40 (n = 131) wherethereis a significant
`benefit from boosters after 1 year (p = 0.003), and
`among patients ages 41 to 55 (n = 125) where
`boosters had no impact (p = 0.97). Patients in the
`18 to 40 and 41 to 55 age groups were also ana-
`
`5
`
`

`

`received boosters (n = 10) and those that did not (n
`= 12). Even in this relatively small group of
`patients, similar findings were observed with these
`patients as with the study group as a whole.
`Specifically, patients with a DSS of 7 who received
`boosters did better than those who did not (p =
`0.04 at 24 months). The effect was again seen pri-
`marily in younger patients (n = 15; p = 0.04 for
`booster efficacy at 24 months; p = 0.04 at 30
`months) and not in older patients (n = 7; p = 0.26
`at 24 months; all failures at 30 months).
`Finally, to determine whether our decision to
`classify patients who withdrew for medical reasons
`as failures affected our results, we reclassified those
`patients as follows. If a patient had declined 1 point
`on the DSS they werestill treated as a failure; if
`they withdrew while stable or improved, they were
`treated as censored. With this classification we
`found an even moresignificant benefit from boosters
`at 24 months, p = 0.006; at 30 months, p = 0.004;
`and at 36 months, p = 0.024. The survival analysis
`also showed a significant benefit due to boosters
`across time, p = 0.027, andafter 1 year, p = 0.028.
`Toxicities. All patients experienced complete
`scalp alopecia with induction. Fever and neutrope-
`nia (WBC, <700/mm*) treated with antibiotics was
`associated with induction in 29 patients: 17 were
`culture negative, four were blood culture positive,
`four were associated with abscess, two with pneu-
`monia, one viral upper respiratory infection, and
`one with an aseptic urinary tract infection. The
`majority occurred early in the study in thefirst 15
`patients treated on 5 consecutive days on a modified
`regimen at a dose of 700 mg/m?. Subsequently, dos-
`ing of 600 mg/m? was given over 8 days. Induction
`was also associated with the following toxicities:
`urinary tract infections (14), oral ulcers (1), candi-
`dal esophagitis (1), gross hematuria (3), and inap-
`propriate ADH secretion (2). Booster therapy was
`associated with the following toxicities: recurrent
`urinary tract infections (4), chronic low WBC that
`did not recover to 4.0 (7), moderate to severe vomit-
`ing (16), and gross hematuria (1). Approximately
`one-third of patients experienced nausea alone with
`induction and on booster therapy. Menstrual abnor-
`malities occurred in approximately half of the
`women that received induction or boosters. There
`were no deaths or secondary malignancies.
`
`Discussion. The primary purpose of the Northeast
`Cooperative Treatment Group was to determine
`whether booster therapy every 2 months with
`cyclophosphamide at a dose of 700 mg/m? could alter
`disease progression in patients with MS. Although
`the results were not dramatic, there wasa statistical-
`ly significant benefit of boosters in the study group.
`Subset analysis demonstrated a strong correlation
`with age in the response to boosters and an age-relat-
`ed response in nonbooster patients at 18 months.
`Because our purpose wasnot to repeat the 1983
`Boston study, and because of the positive results
`reported with the induction regimen, all patients
`
`May 1993 NEUROLOGY43 915
`
`~ 8 H F
`
`TIME TO TREATMENTFAILURE
`
`Age 18-40
`
`Boosters
`
`“xk
`
`OQ
`°oO
`
`&S8
`
`‘8LS
`Qe
`
`igure 2. Kaplan-Meier survival curves comparing time to
`treatmentfailure in patients receiving cyclophosphamide
`boosters versus patients not receiving boosters. Percentage
`of individuals who were not treatmentfailures are plotted
`versus time. Survival comparisons are shown separately,
`first for patients less than the median age of 41 where
`boosters show a significant benefit (p = 0.003), and then
`for patients age 41 and above where boosters had no
`significant impact (p = 0.97).
`
`lyzed to determine if a subcategory of patients
`responded (eg, ages 18 to 30 versus ages 30 to 40),
`but no differences were found. Patients who had
`recent onset of chronic progression (46% stable or
`improved at 24 months with boosters versus 22%
`without boosters, p = 0.02) responded better than
`those who had progressive disease for greater than
`7 years (29% with boosters versus 27% without
`boosters, p = 0.58). Similarly, patients with early
`onset of progression responded better (39% stable
`or improved at 30 months with boosters versus 14%
`without boosters, p = 0.01) than those with onset
`after age 32 (16% with boosters versus 19% without
`boosters, p = 0.87). Of note is that in these analy-
`ses age was a confounding variable, as response to
`therapy was strongly associated with age.
`To determine whether patients with a DSS of 7
`on entry represented a different pattern of progres-
`sion, these patients were analyzed separately.
`There was a total of 22 patients in this category.
`They were equally divided between those that
`
`6
`
`

`

`received induction therapy to maximize the poten-
`tial beneficial effects of boosters. This also allowed
`the physician to be blinded throughout the study.
`The beneficial effects of booster treatment in the
`entire study group, although statistically signifi-
`cant, were relatively modest. We had hoped for a
`stabilization rate of greater than 50% at 2 years
`with boosters. There was, however, a strong associ-
`ation of stabilization with age, and younger
`patients had a stabilization rate that remained at
`40% through 30 months. These findings confirm
`previous reports both by members of our group?
`and Hommeset al® that age is an important vari-
`able in respondingto therapy.
`Because all groups received induction therapy,
`the current study cannot be directly compared with
`the 1983 study in which the primary effect was a
`temporary halt in progression when the cyclophos-
`phamide/ACTH group was compared to an ACTH-
`alone group. The degree to which an untreated or
`steroid- or placebo-treated group would have pro-
`gressed compared with the treatment groups in the
`present study is unknown.It is also possible thatif
`patients were left untreated, they may have done
`as well as the treatment groups, or that induction
`made patients worse and that the positive results
`of the booster infusions related to reversal of such
`negative effects, although we view these possibili-
`ties as unlikely.
`Because the side effects associated with IV
`cyclophosphamide boosters made double blinding
`unfeasible, we did not give a placebo booster treat-
`ment. The examining physicians reported that they
`were unable to tell which patients received boosters.
`However, no formal assessment of blinding was
`done, and no attempt was made to blind the
`patients. It is thus possible that the benefit of boost-
`ers related to a “placebo effect” since half of the
`patients knew they were not being treated. Against
`this possibility is the fact that older patients did not
`respondto boosters whereas youngerpatients did.If
`the response to boosters was related to a placebo
`effect of receiving cyclophosphamide, it would imply
`that placebo effects are st