SEARCH

SEARCH BY CITATION

In the last 20 years, the outlook for rheumatoid arthritis (RA) patients has improved substantially. Wilske and Healey (1) initiated the paradigm shift of reversing the therapeutic pyramid. They made the case that the disease should be hit hard in its initial stages, when a window of opportunity (2) might exist to prevent the damage and severe disability of established disease. This prompted the development of several tight control strategies with methotrexate (MTX) as the anchor drug. In addition, glucocorticoids were rehabilitated somewhat (a process still ongoing), and a whole new class of biologic agents was invented, drugs that proved effective even in patients with longstanding and treatment-resistant disease.

We now face the second wave of development. Although RA patients can now expect rapid relief of symptoms, retention of their paid job, and much reduced disruption of their normal life, therapy is still far from perfect. Therapy fails in significant proportions of patients due to either a lack of efficacy or adverse events. The spiraling costs of biologic therapies are straining the health care budget and threaten to put these drugs out of reach for many patients. In this respect, it is very important that drug development continues, especially the development of cheaper, orally administered small molecules.

Spleen tyrosine kinase (Syk) has emerged as an interesting target for therapy (3). It is an intracellular cytoplasmic enzyme that mediates immunoreceptor signaling in a broad range of immune cells (4). Drugs that inhibit tyrosine kinases first appeared for oncologic indications, with imatinib (Gleevec; Novartis) for the treatment of chronic myelogenous leukemia as the prime example. Drugs that inhibit Syk were shown to have strong antiinflammatory properties in rodent experimental arthritis (5), providing the rationale for experiments in RA. It should be noted that Syk probably has many other functions, including tumor suppression, and that drugs that inhibit Syk also suppress other tyrosine kinases. Thus, short- and long-term safety issues will no doubt surface and will need to be addressed.

Fostamatinib (also called R788) proved to be effective in RA patients with active disease despite MTX therapy in two placebo-controlled trials: a 3-month ascending-dose trial in 189 patients (6) and a larger 6-month trial in 457 patients (7). Especially the higher doses of fostamatinib showed strong effects that overcame a high placebo response. However, diarrhea, upper respiratory tract infection, neutropenia, and hypertension were seen in a significant proportion of patients, leading to premature discontinuation in 5–14%, and dose reductions in 14% in the latter trial.

Unmet need is highest in RA patients in whom treatment with biologic agents is currently failing. Therefore, it is disappointing that the results of a 3-month trial published in this issue of Arthritis & Rheumatism suggest that fostamatinib is not active in this subpopulation of patients with relatively treatment-resistant disease (8). And then the tribulations begin.

Of course, such a disappointing result calls for careful scrutiny of the data to try to explain the discrepancy with the positive results of earlier trials. Such scrutiny usually entails post hoc subgroup analyses, of which Stuart Pocock, a well-known statistician, once stated in a presentation, “Enjoy the results of your subgroup analyses, for you may never see them again.” Pocock et al (9) describe how tests for interaction between a subgroup factor and the outcome are much weaker than tests for the main effect (making it hard to uncover such interaction when it is really there) but also note that post hoc analyses performed after a look at the data frequently show surprising findings and relationships, which on repetition, usually turn out to be spurious. So in the case of a properly designed trial (where the protocol includes procedures to prevent imbalance after randomization), the safest interpretation is that given by the primary analysis. In this case, to quote Genovese et al (8), “There were no meaningful differences seen in the primary end point between the treatment … and the placebo arm”; and this “… calls into question the potential for clinical applicability in this population.”

Nevertheless, the authors uncovered interesting patterns in the data that do cast some doubt on the eligibility of some patients entered into the trial and on the success of randomization in creating prognostically similar groups at baseline. To start with the latter, despite equal levels of disease activity, patients in the active treatment arm had disease that failed to respond to a greater number of biologic agents, had greater prednisone use, and had higher synovitis, osteitis, and erosion scores on magnetic resonance imaging (MRI). The significance of these MRI results is unclear, but the set of findings give some support to the idea that patients in the active drug group may have had disease that was more therapy resistant than those in the placebo group. Confusingly, another post hoc analysis suggested that patients with a higher MRI synovitis score were more likely to meet the American College of Rheumatology 20% improvement criteria (achieve an ACR20 response), which then would “compensate” for the imbalance at baseline.

The authors also argue that the placebo response was inadvertently high, potentially obscuring the effect of active treatment. I find this argument less persuasive: a similarly high placebo response rate was used to calculate the sample size for this trial, and the level was very close to the placebo responses in the other trials, where the drug showed clear separation from placebo. The high placebo response does suggest that this patient group had disease that was less resistant to therapy than expected, which may have been caused by the inclusion criteria that allowed entry into the trial after only 3 months of therapy with biologic agents.

Finally, the authors note high placebo responses in patients admitted into the trial on the basis of a high erythrocyte sedimentation rate (ESR) only, but not in patients admitted on the basis of a high C-reactive protein (CRP) level. The data show that the contrast was actually reversed in these patients, with those taking placebo doing much better than those taking the active drug. This is a very strange result that defies easy explanation, and it could also be a fluke. Perhaps in this group, there was preferential dropout in the active treatment group because of toxicity; such cases would be classified as nonresponders, driving down the results for the active drug group. It would be interesting to see whether a similar pattern exists in the other trials with this agent. In any case, patients with high ESR but normal CRP levels can have severe RA, and there are no data to suggest that such patients respond differently to therapy. Certainly, such patients should not be excluded from trials.

In conclusion, in RA the balance of benefit and harm of Syk kinase inhibition with fostamatinib is currently uncertain. In the MTX failure trials, the drug was effective, but showed substantial toxicity (much more than biologic agents). In patients in whom treatment with biologic agents has failed, the null hypothesis of no effect has not been rejected. Future trials that include such patients should require at least 6 months of prior treatment with biologic agents and should stratify for factors associated with poor response/severe disease, such as glucocorticoid use and prior use of disease-modifying antirheumatic drugs and more than one biologic agent.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. AUTHOR CONTRIBUTIONS
  3. REFERENCES

Dr. Boers drafted the editorial, revised it critically for important intellectual content, and approved the final version to be published.

REFERENCES

  1. Top of page
  2. AUTHOR CONTRIBUTIONS
  3. REFERENCES