A randomized, single-blind study of AZA and prednisone as compared to MTX and prednisone was planned to recruit 60 patients who had recently been diagnosed with generalized MG. This sample size calculation was based on the proportion of patients with MG who were steroid-independent after receiving AZA vs. placebo for a period of 36 months (AZA-Placebo Δ = 40%) [3] and AZA vs. MTX at six months in inflammatory bowel disease (AZA-MTX Δ = 7%) [14], as well as taking into consideration anticipated feasibility and recommendations by Barohn et al. [15].
To meet the objective of comparing the steroid-sparing effectiveness between AZA and MTX, the primary outcome was the average prednisone dose required in each group to achieve and maintain minimal manifestation status (MMS) over 24 months. The clinical MG measurements were assessed by blinded assessors and used to adjust prednisone dosing per protocol, aiming for MMS as defined by the MGFA (i.e. no symptoms of limitation of functioning during everyday living even if some fatiguable weakness was noted on examination) [16]. The MG activities of daily living (MG-ADL) questionnaire was used to objectively assess functional status and subjects in MMS were required to have no MG symptoms (MG-ADL = 0) [17].
Secondary outcomes included the improvement in quantitative MG score compared to baseline, frequency of adverse events and treatment failures in each group. Treatment failures were defined as: withdrawal from the study whether due to intolerance of study medication or uncontrolled MG disease requiring alternative medications; hospitalization for MG relapse or intravenous antibiotic therapy; and death (all-cause mortality).
Subjects
All patients had newly diagnosed (within the previous six months) generalized MG (MGFA class II, III or IV)[15]. A clinical diagnosis of MG had to be supported by at least one of the following: a positive acetylcholine receptor (AChR) antibody assay (AChR-Ab-positive), positive intra-muscular neostigmine test or > 10% decrement on 2-Hz repetitive nerve stimulation [18]. Subjects were required not to have been treated with steroid-sparing agents although they may have been prescribed prednisone prior to inclusion.
For inclusion the severity of MG was such that it resulted in functional disability with respect to the normal activities of daily living despite pyridostigmine therapy so that immunosuppression was indicated. Subjects with AChR-Ab-negative MG and thymoma-associated MG were included if they fulfilled the above criteria and thus required immunosuppressive therapy to control myasthenic disease. Exclusions comprised those with MG confined to the extraocular muscles or concomitant illness such as uncontrolled thyroid disease or additional polymyositis, or subjects with hepatitis B or HIV infections. Thymoma-MG subjects were stabilized on prednisone (± plasma exchange (P/E) or intravenous immune globulin (IVIg)) and surgery scheduled to be performed at the earliest possible time. Due to the concomitant recruitment of generalized AChR-Ab-positive MG patients for a thymectomy trial [19], all such patients opting for this study had refused thymectomy. Potentially child-bearing females were encouraged to use contraception. All patients were advised to refrain from alcohol use [20].
Patients entered after written informed consent and were randomized to either the AZA- or MTX-group using a computer-generated random number sequence. The principal investigator (PI)(JMH), patients and pharmacists were unblinded, but the outcome assessors (AR, KB and RR) remained blinded to the medication subjects were receiving. During the consent process (pre-randomization) it became evident that some subjects could not be randomized as they were unable to afford AZA. Due to limited study funding subjects received their assigned treatment from either the state pharmacy if indigent or from a private pharmacy if they had comprehensive medical insurance. However, participants with partial health insurance, such as a hospital-plan, were not eligible for free state-sponsored chronic medication and could not afford the more expensive AZA. These subjects opted out of randomization but agreed to follow study protocol taking MTX.
The study was approved by the University of Cape Town Health Science Faculty research ethics committee (HSF-067/2005), registered with the South African National Clinical Trials Registry (http://www.sanctr.gov.za) DOH-27-0411-2436, and performed in compliance with the Helsinki Declaration.
Treatment protocol
The dose of AZA approximated 2.5-3 mg/kg daily and MTX 17.5 mg weekly. AZA was initiated at 50 mg daily for 5 days followed by an immediate dose escalation to approximately 2.5-3 mg/kg daily. AZA doses were adjusted in subjects gaining weight. MTX was started at 7.5 mg weekly and escalated weekly by 2.5 mg until a dose of 17.5 mg weekly. The MTX-group also received 5 mg folate for 5 days of every week (25 mg weekly) excluding the MTX dosing day and the day thereafter. Adverse events or toxicity were managed by reducing MTX by 2.5 mg weekly and AZA doses by 25 mg daily.
Medications were scripted by the PI. Prednisone dosing was adjusted according to protocol with the main objective to obtain MMS as recorded by the blinded assessors. Prednisone was initiated at 20 mg daily and escalated by 5 mg weekly until either a dose of 60 mg daily or 1 mg/kg was reached, or the patient reached MMS on a lower dose. Patients in whom prednisone had been initiated prior to study entry underwent a similar prednisone dose escalation from the entry-level dose. Hospitalized subjects underwent more rapid escalations as required. Prednisone tapering (scripted by the PI) was initiated at 5 mg decrements every month once MMS was reached, or if prednisone-related side effects intervened.
Pyridostigmine doses ranged between 180-360 mg daily depending on symptoms. The dose required at the final visit was noted. Vitamin D and Ca2+ supplements were prescribed as is standard practice with prednisone. Patients experiencing dyspepsia were prescribed histamine-2 receptor antagonists (cimetidine) and, if symptoms persisted, gastroscoscopy was performed and a proton-pump inhibitor (omeprazole) initiated (standard care).
MG relapses were managed by increasing the dose of prednisone or, if severe (impending MG crisis), additional immune-suppression was administered such as P/E or IVIg. If an unscheduled visit was required, the PI arranged for staff unrelated to the study to assist in managing the patient.
Visits were scheduled at baseline-, and 1-, 2-, 4- and 6-months after study entry followed by 3-monthly visits for 2 years. At baseline the PI determined the duration of symptoms prior to the diagnosis of MG and before study entry. Pre-study cumulative daily prednisone dose and duration were documented. Patients underwent a full examination, laboratory screening including a complete blood and differential count, urea and electrolytes, liver transaminases, thyroid hormone, random serum glucose and hepatitis B and HIV serological status. A chest X-ray was performed followed by CT scanning of the mediastinum. Subject weight and blood pressure were measured at each visit and the following laboratory tests were monitored: blood counts, serum aspartate aminotransferase (AST), alanine aminotransferase (ALT), gamma glutamyltransferase (GGT) and random glucose. An alcohol intake questionnaire was completed at 6-monthly intervals according to a 3-point scale; 0 did not drink, 1 did some drinking (< 2 drinks per week), and 2 moderate drinking (1 drink most days) [21].
At each visit the blinded assessors performed the MG-ADL and the quantitative MG score [16]. The quantitative MG score was used as described with a minor modification related to hand grip. We anticipated a larger proportion of older subjects and, therefore, adjusted the normal values for the grip dynamometer from those expected for 50-year olds to normal ranges expected for 70-year old participants as suggested by the manufacturer (T.K.K. 5401, Takei Scientific Instruments). The categories were reduced by 66% to accommodate mild, moderate or severe and the value for left-handed grip was 20% below the right hand as recommended by Jaretzki et al. [16].
The PI reviewed medication related side effects and laboratory values. The AZA or MTX doses were adjusted if required. Laboratory toxicity was defined as white cell count ≤3.0 x109/L, neutrophils ≤1.5 x109/L, or platelets ≤100 x109/L, or AST, ALT and GGT > 2-fold the upper limit of normal.
Data management & Statistics
Outcomes related to prednisone dosing, quantitative MG score and MG-ADL were analyzed in as per-protocol analysis in which data measurements of those who withdrew were censored after the date on which they were no longer on study medication. However, data of all subjects were included in the denominator for proportionate outcomes such as MMS and number of failures and "responders". As MG could relapse with prednisone weaning, only those reaching sustained MMS defined as MMS until the end of the study at 24 months, was deemed relevant. The first treatment failure per patient was recorded in the proportional analysis (per 6-month period).
The quantitative QMG scores were modeled as ordinal variables and analyzed using a non-parametric test [22]. Quantitative MG score improvements from baseline (or diagnosis whichever worse if P/E or IVIg were administered after diagnosis but prior to baseline) were calculated by subtracting the value from those at subsequent study visits.
Patients were questioned at each visit regarding compliance. With non-compliance, the recorded dose reflected that which was taken if different from the prescription dose. Patients who missed an appointment were contacted telephonically and the doses taken and MG functional status (MG symptoms) were recorded.
Additional assessments included the number of episodes of worsening during the first year with worsening defined as > 20% deterioration in the quantitative MG score compared to baseline and the proportion of "responders" defined as quantitative MG score improvements of > 3.5 units from baseline [1, 15].
For the statistical analysis, normally distributed data were presented as means with standard deviations (SD) and non- normally distributed data as median and interquartile ranges (IQR). The Student t-test, the Mann-Whitney test and the χ2 test (or Fisher exact test) were used to compare data, as appropriate. All analyses were two-sided and a p-value < 0.05 was considered significant. Statistical analyses were done using STATISTICA 9 (Statsoft®).