Treatment with the phosphodiesterase type 4 inhibitor rolipram fails to inhibit blood brain barrier disruption in MS

Abstract

Rolipram, a prototypic phosphodiesterase-4 inhibitor is highly effective in suppressing Th1 autoimmunity in multiple animal models, including experimental autoimmune encephalomyelitis (EAE). Additionally, Rolipram has been extensively studied as a potential neuroprotective agent. Based on its anti-inflammatory activity, we tested the efficacy of Rolipram in suppressing inflammatory disease activity in MS in a proof-of-principle phase I/II open label clinical trial. Enrolled MS patients were evaluated by monthly MRI and clinical examinations during three months (four MRIs) of pre-treatment baseline and eight months of Rolipram therapy. The primary outcome was the change in contrast-enhancing lesions (CEL) between baseline and last four months of Rolipram therapy. Previously defined biomarkers of Rolipram-mediated immunomodulation were evaluated during study. The trial was stopped prematurely because the drug was poorly tolerated and because of safety concerns: we observed an increase, rather than decrease in the brain inflammatory activity measured by contrast-enhancing lesions (CEL) on brain MRI. At the administered doses Rolipram was active in vivo as documented by immunological assays. We conclude that the reasons underlying the discrepancy between the therapeutic efficacy of Rolipram in EAE versus MS are at present not clear.

Introduction

Cyclic nucleotides, especially cGMP and cAMP, are important as second messengers in cell signaling. In the cells of the immune system, increased intracellular levels of cAMP activate Protein Kinase A (PKA), which is a potent inhibitor of the immune system [1]. Intracellular cAMP levels are regulated by G-protein-coupled receptors, adenylyl cyclases (AC) and phosphodiesterases (PDEs). In immune cells, cAMP is cleaved predominantly by phosphodiesterases 4 (PDE-4) and PDE-3 [2, 3]. Therefore, agents that block PDE-4 increase intracellular cAMP levels and subsequently modulate immune functions such as T cell activation and –cytokine secretion [3–6].

One well studied PDE-4 inhibitor is rolipram; a drug initially developed based on its antidepressant properties. In-vitro inhibition of PDE-4 by rolipram decreases production of Th1-like cytokines such as TNF-α and INF-γ by activated human T cells [5–7]. Consistent with these changes, studies of PDE-4 inhibitors in several models of experimental allergic encephalomyelitis (EAE) have shown a beneficial effect on disease activity [6, 8–14]. In addition, PDE-4 inhibitors ameliorated disease in a variety of other experimental autoimmune diseases such as experimental autoimmune uveitis [15], collagen-induced arthritis [16, 17] and diabetes in the NOD mouse [18]. In all these animal models, T cells of Th1 or, as documented much later, Th17 phenotype underlie immune-mediated tissue destruction. Based on observations indicating that Th1 cells can be encephalitogenic in humans [19] we reasoned that PDE-4 inhibitors such as rolipram might block inflammatory disease activity in MS.

The current report describes the results of a phase I/ early phase II clinical study of rolipram in MS. As will be described the study was terminated early because of tolerability and lack of efficacy.

Materials and Methods

Clinical trial design

Rolipram raw material, as well as toxicity and pharmacology data were made available to the principal investigator of the study (R. Martin), Cellular Immunology Section, Neuroimmunology Branch (NIB) of the National Institutes of Neurological Disorders and Stroke (NINDS), NIH, by Schering AG, Berlin, Germany, under a Collaborative Research and Development Agreement (CRADA). Recrystallization, regulatory application for an FDA IND (# 60,932), and part of the trial were supported by an NINDS intramural Bench-to-Bedside grant. Rolipram was re-crystallized from the raw material under GMP conditions by an independent contractor and packed into capsules by the NIH pharmacy.

The main purpose of this open label, baseline-versus-treatment clinical trial was to test the safety, tolerability and effects of rolipram on brain inflammatory activity in MS patients. The trial consisted of two stages with identical trial design (Figure 1): Stage I was a safety- and dose-finding study in 6 patients with definite relapsing-remitting (RR-) or secondary-progressive (SP-) MS [20] with moderate disability (expanded disability status scale (EDSS [21] 4.0–6.5; EDSS of 0 meaning without neurological deficit and EDSS of 10 meaning death from MS), who had no or low brain inflammatory activity. These patients are identified by codes R101-R106.

Figure 1. NIB rolipram trial design.

A/ Schematic delineation of the rolipram trial design. Shaded areas represent predetermined comparison for the primary outcome measure. PK = pharmacokinetics. Triangle under “drug dosing” represents slow pre-determined titration of rolipram during the first month of rolipram therapy.

The stage II study (initiated after 4/6 patients in Stage I finished the dose-escalation phase, and safety and tolerability data were collected) stipulated up to 20 RR- or SP-MS patients with EDSS 1.5–6.5 and moderate brain inflammatory activity (average at least 0.5 CEL per scan during 4 monthly pre-treatment MRI scans). Only two patients were enrolled and completed dosing (R201-R202) before the trial was terminated.

Enrolment criteria included that patients had to be off treatment at baseline, or could not tolerate FDA-approved disease-modifying therapies, or chose not to be treated or their neurologist referred them to the trial because they had not responded to prior treatment. Two out of 8 enrolled patients were treatment-naïve, while remaining six received several trials of disease modifying therapies, including IFN-β, copaxone and monthly infusions of solumedrol. However, all patients were untreated for at least 1 year prior to enrollment into current trial. Both groups underwent rolipram dose escalation during the first 28 days according to a pre-determined dosing schedule (Figure 1). Patients in Stage 1 were dosed up to 9mg/day (3 mg, 3 times daily); however, only 2/6 patients could be dosed beyond 7.5mg/day and dose of one of these patients had to be decreased back to 7.5mg/day because of development headache and fatigue. Consequently, patients in Stage 2 were dosed only up to 7.5mg/day, which was selected as highest tolerable dose based on the results of Stage 1. Patients were followed by monthly MRI and clinical examinations (Figure 1 and Table 1). The primary outcome measure was inhibition of brain inflammatory activity measured as contrast-enhancing lesions (CEL) on brain MRI during the last four months of rolipram therapy (Months 5–8) compared to the four months of pre-treatment baseline (Months -3 to 0). Three types of analyses were predetermined by the protocol: analysis of individual subgroup (i.e. Stage I and II patients separately) and joint analysis of all patients (i.e. Stage I and II together). Secondary outcome measures were changes in volumetric MRI parameters (see MRI methodology) and clinical parameters (EDSS [21], Scripps NRS [22] and MS functional composite score (MSFC)[23]) as defined in Table 1, comparing the same treatment periods.

Table 1.

Adverse events and clinical/MRI outcome measures

	Baseline	Rolipram Therapy	Statistical significance
	33 patient-months1	55 patient-months2
	2.75 patient-years	4.58 patient-years	P value
Adverse events
All adverse events	10 (3.63 per patient/year)	37 (8.08 per patient/year)	P = 0.01
Infections	2 (0.73 per patient/year)	9 (1.97 per patient/year)	ns
MS exacerbations	3 (1.09 per patient/year)	6 (1.31 per patient/year)	ns
Clinical data
EDSS3 (Median, Mean/SD)	5.3 (4.6/1.9)	5.3 (4.5/2.1)	ns
Scripps-NRS4 (Median, Mean/SD)	63.4 (67.0/12.6)	63.7 (71.6/13.2)	ns
MSFC5 (Median, Mean/SD)	−0.043 (−0.037/0.788)	0.124 (0.030/0.760)	ns
MRI data
New CELs/month (Median, Mean/SD)	0.38 (0.36/0.33)	0.63 (1.17/1.50)	ns
Predetermined analysis6
New CELs/month (Median, Mean/SD)	0.25 (0.30/0.31)	0.35 (0.66/0.94)	ns
All patients/entire treatment period
Total CELs/month (Median, Mean/SD)	0.44 (0.58/0.36)	1.00 (1.71/2.19)	P = 0.052
Predetermined analysis6
Total CELs/month (Median, Mean/SD)	0.38 (0.38/0.34)	0.51 (0.95/1.33)	ns
All patients/entire treatment period
Volume of CELs/month (Median, Mean/SD)	0.008 (0.035/0.074)	0.023 (0.154/0.293)	ns
Predetermined analysis6
Volume of CELs/month (Median, Mean/SD)	0.008 (0.028/0.064)	0.016 (0.069/0.137)	ns
All patients/entire treatment period
T2LL7/month (Median, Mean/SD)	1.752 (1.795/1.109)	1.636 (1.791/1.159)	ns
Predetermined analysis6
T2LL7/month (Median, Mean/SD)	2.051 (2.432/1.967)	1.930 (2.363/1.840)	ns
All patients/entire treatment period
BPF8 (Median, Mean/SD)	0.8371 (0.8397/0.0145)	0.8390 (0.8425/0.0148)	ns
Predetermined analysis6

¹Four months of untreated baseline per patient; baseline period was extended by one month if i.v. solumedrol was used for the treatment of MS exacerbation

²Eight months per patient, but only four months for R105 and three months for R106

³Expanded Disability Status Scale [21] from 0 (normal exam) to 10 (death due to MS)

⁴Scripps Neurological Rating Scale [22] from 100 (normal exam) to 0 (death due to MS)

⁵Multiple Sclerosis Functional Composite [23]

⁶Predetermined analysis: compares four months of baseline with last four months (Mo 5–8) of rolipram therapy. Only six patients that finished rolipram dosing are included

⁷T2 lesion load (nonspecific measure of diseases brain tissue)

⁸Brain Parenchymal Fraction (measure of brain atrophy)

The trial design was approved by the NINDS Institutional Review Board (IRB), and all patients signed the informed consent. Safety and tolerability was assessed by an independent data and safety monitoring board (DSMB).

MRI collection and analysis

Contiguous interleaved axial MRI images (3mm × 42 axial slices) were acquired at 1.5T (Signa, GE Medical Systems, Milwaukee, WI) with a standard head coil using a FOV (24×18cm) and matrix (256×192). Pulse sequences included: Dual echo, FSE T2 (TR:TE 3400/17/104 msec), fluid attenuation inversion recovery (FLAIR; TR; TE:TI 10,000, 140, 2200msec)-, and T1-weighted (TR; TE 600; 16 msec) before and after contrast (Magnevist 0.1 mmol/kg; Berlex Laboratories, N.J.) administration as described [19]. CEL were recorded on hard copy films by consensus of two radiologists. T2LV was determined by a semi-automated thresholding technique (PV-WAVE) [24]. The volume of CEL was determined from registered images [25] using a semi-automated thresholding program (Jeff Solomon, MRIPS, NIH) on MEDX (Sensor Systems, Sterling, VA) applied to post-contrast T1WI, after verification of lesion co-localization on T2WI or FLAIR images. Brain parenchymal fraction (BPF) was determined on registered, pre-contrast T1W images using the automated program (ADPK mean) as previously described [26].

Prospective immunophenotyping by flow cytometry

Blood samples (10 cc of EDTA anti-coagulated blood) were collected bi-monthly between 8:30–11:30am and were processed within 2h. Surface markers were evaluated from fresh cells prospectively by 3–4 color flow cytometry using commercially available fluorescent antibodies (Ab) after RBC lysis as described [27]. The absolute numbers of CD4+ and CD8+ T cells and CD19+ B cells and CD14+ monocytes were calculated from application of percentages of these cells derived from flow-cytometry analyses to absolute numbers of lymphocytes, determined by the NIH Clinical Center laboratory from parallel whole blood samples.

Additional assays from cryopreserved PBMC

Lymphocytapheresis was collected during baseline, at month three and eight of rolipram therapy between 8:30–11:30am. PBMC were isolated by density gradients within 2h of ex-vivo collection and were cryopreserved.

For monocytes and B cell activation assay, PBMC from baseline and rolipram therapy samples were thawed and processed simultaneously. B cells (CD3-/CD19+ lymphocytes) and monocytes (CD14+) in thawed PBMC cultures were stained for the surface expression of CD80 and CD86 immediately after thawing and 20h after activation with LPS (1μg/ml) and analyzed by flow cytometry. Upregulation of CD80 expression on activated B cells and monocytes was compared between baseline and therapy samples.

For CFSE proliferation assay, PBMC were stained with 5-(and-6)-carboxyfluorescein diacetate succinimidyl ester (CFSE; Invitrogen; 1μM) as described [28]. After washing, PBMC at a density of 1×10⁶ PBMC/ml were activated with plate-bound anti-CD3 (20ng/ml; pre-incubated at 37°C for 2h) and CD28 (CD28 at 10μg/ml). Six days post-stimulation, cells were stained for surface population markers and analyzed by flow-cytometry. We calculated the total number of mitoses per 100-gated cells using formula: # of mitoses = Σ (X_n × 100 – X_n × 100/2ⁿ) where X is the percentage of cells that underwent n-divisions.

Statistical analysis

Changes in clinical and MRI measures between baseline, treatment and post-treatment were assessed by Friedman Repeated Measures Analysis of Variance on Ranks. For categorical data (adverse events) Fisher’s exact test was used. In order to evaluate immunological changes from baseline to treatment, 2 baseline samples and 2–3 treatment samples were averaged for each individual and compared by Signed Rank Test or, if permitted, by paired T-test. All statistics were performed using Sigma-Stat 3.5 (Jandel Scientific, CA), with a pre-set limit of statistical significance of P<0.05.

Results

Safety and tolerability: Rolipram therapy was poorly tolerated, but it did not lead to significant changes in clinical measures of disability

During the pre-treatment baseline we observed a total of 10 adverse events (AE; 3.63 per patient-year; Table 1); all of them grade 1–2 (NCI Common Terminology Criteria for Adverse Events; http://ctep.info.nih.gov). The total number of AE increased to 8.08 per patient/year during rolipram therapy (223%; P=0.01, Fischer’s Exact Test); all of them were grades 1–2 except two, which were grade 3 (atypical hypersensitivity reaction to rolipram in patient R105, which lead to discontinuation of rolipram after four months of treatment, and severe insomnia, gastroesophageal reflux and balance changes in patient R106 resulting in her withdrawal from the trial after three months of treatment). No difference in the frequency of AE was noted between patients who received 9mg/day (2/8) versus those that received 7.5mg (6/8) of rolipram per day. The number of infections increased from 0.73 per patient-year during baseline to 1.97 per patient-year during rolipram therapy (269%; not significant). We observed 3 exacerbations during baseline (1.09 per patient-year) and 6 exacerbations during rolipram treatment (1.31 per patient-year). The remaining adverse events observed during rolipram therapy were most commonly nausea, vomiting and insomnia.

We saw no significant changes in the clinical measures of disability; they all remained stable or slightly improved (Table 1).

Primary outcome measure: Rolipram therapy increased the number of contrast-enhancing lesions (CEL) on brain MRI

Besides the poor tolerability the main concern of the investigators and the DSMB was the increase in CEL observed during rolipram therapy (Table 1), which was especially evident during the latter half of the treatment period (Figure 2C). According to pre-determined analysis (i.e. comparing four months of pre-treatment baseline with the last four months of rolipram therapy (Figure 2C, darker shaded area: Months 5–8; only six patients who concluded the eight months of dosing are included) the total number of CEL per patient per month increased from a median of 0.44 to a median of 1.00 (227.3%; P=0.052) at the time when the investigators in consultation with the DSMB decided to terminate the trial early. As evident from the data of individual patients (Fig 2A and B), the total number of CEL increased during the last 4 months of rolipram therapy in 3/6 patients in Stage I (the remaining 2/6 patients stopped rolipram dosing before Month 5 and 1/6 patients had no CEL during entire trial) and in 2/2 patients in Stage II as compared to pre-treatment baseline periods.

Figure 2. Effect of rolipram on brain MRI contrast-enhancing lesions (CELs): Primary outcome measure.

Number of new (black circles) and total (open triangles) CELs in individual patients from A/ Stage I (low inflammatory MS) and B/ Stage II (high inflammatory MS) of the trial. C/ Cumulative number of contrast-enhancing lesions (CELs) per entire cohort per month. Duration of rolipram treatment is highlighted by gray shading. Pre-determined analysis compares the number of total and new CELs during four months of baseline (Months -3 to 0) and the last four months of treatment (Months 5–9; darker gray shading).

The volume of CEL also increased from a median of 0.008 cm³ to 0.023 cm³ (287.5%). T2 lesion load (T2LL) values remained unchanged. Although we observed a non-significant increase in Brain Parenchymal Fraction (BPF) during rolipram therapy, this was more likely due to edema associated with the increased CEL than due to stabilization or reversal of brain atrophy.

In order to provide data on the entire cohort, Table 1 includes in addition to predetermined analyses also MRI and clinical data for all patients during the entire treatment period.

Ex-vivo immunophenotyping of B cells, T cells and monocytes

In our in-vitro studies [5] we identified several effects of rolipram on the human immune system, which we used as biomarkers to monitor in vivo biological activity of rolipram during the current trial. Among these were the induction of CD86 expression on B cells and inhibition of CD80 expression on LPS-activated B cells and monocytes. By ex-vivo immunophenotyping performed prospectively every two months during the trial, we observed a statistically-significant increase in the percentage of CD86+ (but not CD80+) B cells during rolipram therapy (Figure 3A). Similarly, we observed a statistically-significant inhibition of surface CD80 upregulation on activated B cells and monocytes during rolipram therapy (Figure 3B).

Figure 3. Effect of rolipram therapy on CD80 and CD86 expression on B cells and monocytes.

A/ Prospective ex-vivo immunophenotyping of CD80 and CD86 surface expression on CD19+ B cells. For each subject, baseline data-point represents average of two baseline samples and therapy data-point represents the average of 2–3 therapy samples acquired prospectively, every other month. B/ From cryopreserved PBMC samples monocytes and B cells were activated with LPS. Surface expression of CD80 on CD19+ B cells and on CD14+ monocytes was evaluated before and 20h after activation by flow cytometry. C/ Changes in subpopulations of immune cells induced by rolipram therapy. For each subject, baseline data-point represents average of two baseline samples and therapy data-point represents the average of 2–3 therapy samples acquired prospectively, every other month.

Additionally, we observed changes in the prospectively acquired immunophenotyping data that were not predicted from our in-vitro experiments: we observed a mild, but statistically significant decrease in the percentage and absolute numbers of monocytes and CD4+ T cells during rolipram therapy (Figure 3C), although none of these values fell outside the normal range.

Inhibition of T cell proliferation ex-vivo during rolipram treatment

Our previous in-vitro experiments also demonstrated that rolipram, at the concentrations achievable in-vivo (i.e. 0.1μg/ml) inhibits T cell proliferation [5]. Therefore, we assessed T cell proliferation to polyclonal stimuli in cryopreserved PBMC derived from baseline, Month 3 and Month 8 of rolipram therapy samples without exogenous addition of the drug. We observed a statistically significant inhibition of CD4+ and CD8+ T cell proliferation, which was more robust at a later therapy time-point (Month 8).

Discussion

Rolipram, a prototypical PDE-4 inhibitor, represented a highly attractive therapy for MS based on several criteria: First, it had beneficial therapeutic effect on multiple EAE models, both in prevention and treatment paradigms [6, 8–14]. Second, it was later also shown to have neuroprotective properties [29–31], and finally, at the time when this trial was initiated, it represented one of the few oral agents with therapeutic potential for MS [32]. Despite the wealth of promising data the present proof-of-principle clinical trial of rolipram in MS had to be stopped based on the poor tolerability of rolipram and the lack of efficacy or possibly even increase in inflammatory activity as measured by MRI. When the changes in total CELs/month during the 4 months of baseline were compared to the last 4 months on therapy an increase during the treatment phase was noted, and despite the small number of patients this increase nearly reached statistical significance (225% increase; P=0.052). An increase in the number and/or volume of CEL during rolipram treatment was observed in six of eight individuals (the two remaining individuals had no inflammatory lesions during the entire trial), indicating that the group change was unlikely to be driven by a single or few outliers. Also, in two patients with active disease (Stage II patients) the level of activity appeared greater in the treatment phase than would have been predicted from the level of activity during the baseline evaluation. With the type of trial design that was used here, it is expected that, if the drug has no effect on MS inflammatory activity, the number of CEL will remain stable or even decrease (due to regression to the mean) during treatment as compared to baseline. This notion is underlined be our prior experiences with an identical trial design. We never observed an increase in CELs during treatment in eight different trials with five different therapies at the group level ([19, 33–36] and unpublished observations). In view of these data, the observed increase in CELs during the therapeutic phase of this trial was concerning, even though it was not accompanied by clinical deterioration. However, because of the small number of patients, it is impossible to conclude conclusively that rolipram increases inflammatory disease activity in MS. Also consistent with an accentuation of disease is the small increase in the number of exacerbations adjusted for exposure. However, again the small number of patients requires that these changes be regarded with caution.

What could explain a lack of effect or even an increase in inflammatory activity in MS when previous studies in EAE have shown a prominent anti-inflammatory action of the drug? Our immunological data clearly indicate that rolipram was pharmacologically active in-vivo, and the observed increase of CD86 expression on resting B cells as well the decrease of CD80 expression on activated B cells and monocytes is consistent with our prior in-vitro studies. Finally, we observed a reduced proliferative capacity of CD4+ and CD8+ T cells derived from rolipram therapy samples as compared to pre-treatment PBMC samples. All of these biomarker analyses were predetermined and based on our prior data on the in-vitro effects of rolipram on human immune cells [5]. We observed two additional changes that were not predicted from in-vitro studies. Rolipram treatment resulted in a mild decrease in blood monocytes and CD4+ T cells. Since rolipram is highly lipid soluble and rapidly passes the blood brain barrier (BBB), it is expected that its immunomodulatory actions will also occur in the CNS compartment. The observations from the current trial show that, despite the many promising findings in animal models and in-vitro studies, inhibition of PDE-4 does not block inflammatory disease activity in MS, which is considered a prototypic Th1-mediated autoimmune disease. We currently do not know if this discrepancy is due to profound differences in the pathogenesis of EAE versus MS as has been suggested by other failed translations from the model to the human disease or if we did not sufficiently consider factors other than the Th1/Th2 balance such as regulatory immune cells such as Foxp3+ T cells [37], IL-10-producing T cells [38] and other regulatory populations such as CD56^bright NK cells [28].

Despite the apparent lack of efficacy of rolipram on acute inflammatory activity in MS, the possibility that the drug could contribute to neuroprotection remains [29–31]. Furthermore, PDE-4 inhibitors have been shown to have immunomodulatory effects that may target events in the MS lesion other than those contributing to blood brain barrier disruption seen on contrast-enhanced MRI, such as e.g. production of nitric oxide. The current study did not address these issues, but we can conclude at this point that any future use of PDE-4 inhibitors in the treatment of MS must be approached with caution and using studies that are carefully monitored for increase in disease activity. Our experience with rolipram raises an additional important point that should be considered in future neuroprotective trials in MS. In contrast to purely neurodegenerative disorders, the use of neuroprotective agents in MS may be complicated by the effects of these agents on the immune system, which may inadvertently enhance MS-related inflammation, thus limiting their overall therapeutic potential.

Figure 4. Effect of rolipram therapy on T cell proliferation.

Cryopreserved PBMC were thawed and analyzed in parallel for all tree time-points. PBMC were CFSE stained and polyclonally activated for 6–7 days. Number of mitoses was calculated independently for CD4+ and CD8+ T cells. Mo3 Th = month 3 on therapy, Mo8 Th = month 8 on therapy.