We thank Dr. Cook and Dr. O'Connor and his colleagues for their interest in our article. Their comments underscore the diagnostic and therapeutic challenges associated with atypical presentations of granulomatous diseases. Dr. Cook insightfully points out that the granulomatous variant of common variable immunodeficiency can be confused for sarcoidosis. Although not presented in our article, immunofixation electrophoresis of our patient's serum in fact demonstrated polyclonal hypergammaglobulinemia, and while serum IgG and IgM levels were not quantitatively assessed, serum IgA levels (determined in anticipation of possible intravenous gammaglobulin therapy for the antiphospholipid antibody syndrome) were normal. These data are thus more consistent with sarcoidosis, which is associated with hypergammaglobulinemia, and less consistent with common variable immunodeficiency, which is characterized by panhypogammaglobulinemia. Nonetheless, we firmly agree with Dr. Cook that the clinical similarities between sarcoidosis and granulomatous common variable immunodeficiency probably reflect common pathophysiologic pathways and therefore suggest potentially overlapping therapeutic approaches. Animal models and human studies implicate TNF- in the pathogenesis of granulomas. Mice deficient in either TNF- or TNF receptor I exhibit impaired granuloma formation in response to bacterial antigens (1, 2); as pointed out by Dr. Cook and in our report, various lines of evidence suggest that active granulomatous disease in humans is associated with increased production of TNF-. It is therefore not surprising that certain features of granulomatous common variable immunodeficiency may respond to TNF- inhibition (3) and that additional cases of refractory sarcoidosis responding to infliximab have been reported by Baughman and Lower (4).

Dr. O'Connor and colleagues warn of the potential exacerbation of occult Mycobacterium tuberculosis infections with anti–TNF- therapies. We would extend this caution to include all types of infection, especially atypical ones such as other mycobacteria or fungi that may be more difficult to identify and may in fact mimic systemic inflammatory disorders. In our patient, corticosteroids and infliximab were not initiated until we were reasonably comfortable that infectious causes were excluded. Postlicensure reports of infections in the setting of infliximab and etanercept therapy have included tuberculosis, listeriosis, Pneumocystis carinii pneumonia, herpesvirus infections, and candidiasis (5). Vigilance and suspicion for active or latent infectious conditions are essential before, during, and after anti–TNF- treatment, as they would be with any immunosuppressive therapy. Moreover, since TNF- may contribute to tumor surveillance and immune tolerance, we further recommend that, with anticipated wider use of anti–TNF- therapies, physicians must also maintain increased awareness of potential oncologic and autoimmune complications.