Whither Methylene Blue (TRx0237) in Alzheimer’s?

Sometime in early January, I starting wondering about the status of TRx0237 (aka methylthioninium chloride (MTC), aka methylene blue) one of the few presumptive disease-modifying molecules (as a tau-aggregation inhibitor) still in late-phase development for established Alzheimer disease (AD). The Singapore-based company, TauRx, had reported the phase 2 results of its AD compound in abstract form in 2008, but there was little in the way of follow-up and certainly no publication of the phase 2 trial results in a peer-reviewed journal. Citing a 2012 Q&A at alzforum with TauRx CEO Claude Wischik, Berkwyk and Sabbagh wrote in 2013,

To date, no data regarding the Phase I and Phase II trials of TRx0237 have been published. The rationale given by TauRX Therapeutics is that Phase II data tends to not be meaningful in terms of ultimate outcome of the clinical process, and that the data regarding safety and efficacy has been presented to regulating authorities in the United States, Europe and Asia has engendered enough confidence to secure a Phase III trial as well as almost entirely private funding for the study.

That may be, but I’ve just now learned that the phase 2 results were indeed published in a peer-reviewed journal in January* and in the context of a somewhat-understated press release from the company. In addition to the curious lack of fanfare that typically goes along with this type of publication, there was almost no press attention to this news. In fact, a search of Google News brought up nothing recent on the company generally or the phase 2 announcement specifically. The lone (discovered) exception was FiercePharma Asia, which last week provided a profile of Wischik and his funding efforts, in the context of a limited discussion of the phase 2 results for TRx0237. I find this publicity desert odd, but it may be explainable given the mixed, and maybe even questionable, results of the study.

Here are my take-aways from the January peer-reviewed article. Notably these results were from individuals with AD who were not actively taking cholinesterase inhibitors (eg, donepezil [Aricept]) or memantine (ie, Namenda).

1. At 24 weeks in patients with moderate AD (modified intent-to-treat population, n = 37), there was a statistically significant difference between the ADAS-cog score in 17 patients taking the intermediate dosage (138 mg/d [60-mg capsules TID]) of TRx0237 and that in 20 patients taking placebo (n = 20) (ADAS-cog score differences: -1.08 with treatment vs 4.35 with placebo). (Note: A higher, positive ADAS-cog score indicates more cognitive impairment. Marked clinical worsening has been defined roughly as a ≥4-point increase in the ADAS-cog score.) The effect size—meaning, in this case, the straight-up difference between the ADAS-cog scores in the 2 treatment groups—was -5.42 units (corrected P value = .047). Statistically significant differences between treatment and placebo were also noted with the secondary outcomes of the MMSE (a common, office-delivered cognitive test) and the ADCS-CGIC (a recognized global-function test). The effect size for the MMSE was 3.79 units (corrected P = .028), which is probably clinically recognizable.

The authors did not provide Cohen’s d effect sizes, a potentially useful tool for comparing drug efficacies across studies. I attempted to calculate them myself, but I suspect my results are unreliable (or at least not very useful) given the small number of people who actually had moderate disease (only about 20% of the study population).

2. In a (presumptive) effort to support the idea that TRx0237 is a disease-modifying drug (as opposed to a merely symptomatic therapy, like a cholinesterase inhibitor^†), the investigators incorporated the biomarker of regional cerebral blood flow (rCBF via SPECT) in a subset of 135 patients. For this secondary outcome, there was statistically significantly less decline in rCBF in patients taking either the intermediate or high dosage of TRx0237 at 24 weeks. However, the biomarker effect was seen only in patients with mild disease, not in moderate disease (which is known to have more advanced perfusion deficits, according to the authors). The rCBF treatment differences were greatest in the bilateral medial-temporal and temporoparietal areas of the brain.

I was unable to calculate Cohen’s d effect sizes for these data, using the Table 3 values, because the n values for the various treatment groups in this imaging substudy weren’t given. But again, these effect-size calculations probably wouldn’t have been all that reliable anyway, because of the suspected small substudy n values. (In the discussion, the authors reveal that there were only 25 treated subjects with mild disease in the rCBF substudy.)

3. In an “exploratory post hoc analysis” of data at 50 weeks (first extension phase, 24-50 weeks), the intermediate dosage of TRx0237 (n = 49) was also associated with significant differences on the ADAS-cog in patients with either mild or moderate disease (vs initially placebo-treated patients with mild disease; n = ??). In essence, TRx0237-treated patients did not show cognitive decline, while initially placebo-treated patients with mild disease did. (More on this group below—see item 6.) In addition, the authors calculated a very strong correlation between the 50-week ADAS-cog effect size in treated patients with mild disease (n = ??) and their 24-week rCBF data (n = ??) (r = 0.989).

Once again, I could not calculate Cohen’s d effect sizes, because a breakdown of the n values in the extension phase for the various substudy groups weren’t given.

4. The most common AEs with TRx0237 were gastrointestinal or renal/urinary. AEs leading to permanent discontinuation of TRx0237 at the 138-mg/d dosage were exceptionally high: 35%. Among these were 4 cases of methhemoglobinemia (>3.5% threshold level for methemoglobin).

5. The lack of any clinical efficacy signal with the highest dosage of TRx0237, 228 mg/d apparently led to a lot of head scratching among the authors. I’m speculating here, but this may be the reason for the delay in the peer-reviewed publication of the phase 2 results.  Suffice it to say that the authors believe that this “entirely unexpected” result is due to absorption challenges, particularly when TRx0237 is taken with food, and hinges on whether the molecule is in the reduced or oxidized state in the gut.

6. There’s a confounding factor in the initially placebo-treated patients (n = 78), who entered the extension phase. According to the authors, this group received a “nominal” dosage of 152 mg/d (100-mg capsules BID) of TRx0237 from weeks 24 to 50; but they received only 73 mg/d of the compound because of “absorption and dissolution limitations.” Therefore they viewed this group as a “relatively inactive control arm” throughout the extension phase. This group, which did not show decline during the first 24 weeks of the study, showed a 2.94-point decline on the ADAS-cog scale at 50 weeks (vs essentially no ADAS-cog decline in the initially treated mild or moderate subjects taking 138 mg/d). The authors concluded (in a very confusingly worded statement, IMO) that these data support a disease-modifying effect for the drug.

7. In addition, there may have been some issues with proper blinding, given that TRx037 causes urine discoloration (methylene blue + yellow = green).

Obviously TauRx and evidently regulatory bodies believed that these phase 2 data were sufficient to support phase 3 development of TRx0237. Specifically the study authors, while acknowledging that “the number of subjects was small in the treatment arm of greatest interest,” proposed that “statistical significance was achieved only because of the large effect sizes observed.”

In phase 3 development, TRx0237 will be delivered in a new “stable reduced form” (LMTX^®), to address any delivery issues in the phase 2 trial. Dosages will range from 150 to 250 mg/d. An international 18-month phase 3 study of the drug (200 mg/d) is ongoing in patients with mild disease. The estimated enrollment is 700 patients with a study completion date of May 2016. A 15-month phase 3 study of 150 or 250 mg/d in 833 patients with mild or moderate disease is scheduled to be completed by February of next year. In both studies, FDG-PET imaging will be the presumptive biomarker to support disease-modifying effects.

ADAS-cog = cognitive subscale of the Alzheimer’s Disease Assessment Scale; ADCS-CGIC = Alzheimer’s Disease Cooperative Study Clinical Global Impression of Change; CI = confidence interval; FDG-PET = fludeoxyglucose (18F)-positron emission tomography; MMSE = Mini-Mental State Examination; SPECT = single-photon emission computed tomography.

*And online in December.

†Although debate goes on about whether there’s a hard-and-fast distinction between a symptomatic therapy and a disease-modifying therapy in AD.