I read with interest the article “Initiation of anti-TNF therapy and the risk of optic neuritis: from the safety assessment of biologic ThERapy (SABER) Study”; however, I have some hesitation about drawing major conclusions from this experience.

Like many serious drug-related side effects surfacing post Food and Drug Administration (FDA) approval, optic neuritis associated with anti–tumor necrosis factor (TNF) therapy is a rare event, and accordingly requires large sample sizes for proper evaluation. The SABER study accomplished this, but equally important in studies with rare outcomes is case ascertainment. How sure can the reader be that a case classified as “optic neuritis” in this retrospective database review was actually optic neuritis and not another optic neuropathy or other cause of visual loss? The screening procedure for optic neuritis in the SABER study employed “a combination of ICD-9 codes for optic neuritis and laboratory tests (eg, rapid plasmin regain [RPR], serum angiotensin converting enzyme [ACE] level, orbital magnetic resonance imaging, others) typically ordered during an acute optic neuritis examination. We identified optic neuritis cases using the following algorithm: any patient given 1 inpatient or outpatient ICD-9 code for optic neuritis (377.30 or 377.32) with evidence of serum ACE testing within 90 days of the code OR any patient given an optic neuritis ICD-9 code 3 times within 60 days.” Prior studies have concluded that International Classification of Diseases, ninth revision (ICD-9) coding is often inaccurate, likely accounting for the author’s use of multiple identification methods. Although the authors report that this search algorithm was “validated” and associated with a positive predictive value of 100%, they do not report what gold-standard definition of optic neuritis was used in this assessment.

Many in the neuro-ophthalmology community (those who are principally involved in nonglaucomatous optic neuropathy evaluation) define optic neuritis as an inflammatory demyelination disease of the optic nerve, based on the Optic Neuritis Treatment Trial (ONTT) definition. It is this definition that is most germane to the issue of TNF inhibitors, typically affecting only 1 eye and excluding other causes such as sarcoid, ischemia, or infection. The Optic Neuritis Treatment Trial concluded that there was no benefit from blood tests in typical optic neuritis; one could argue that the SABER-captured cases were not typical, but certainly serum ACE testing shouldn’t correlate strongly with a final diagnosis of typical optic neuritis.

The primary analysis among current anti-TNF inhibitor users found “optic neuritis” rates of 8.9 and 43 cases per 100 000 person-years in rheumatoid arthritis and inflammatory bowel disease, respectively, compared to rates of 0 for non–TNF inhibitor treatment in these populations—certainly suggestive despite the small numbers. One of the SABER cases of optic neuritis associated with anti-TNF therapy involved an 85-year-old patient—this is just not the age when “optic neuritis” occurs. This case suggests either a causal association with the drug, or a misdiagnosis; however, without knowing case specifics such as examination, course, imaging, and outcome, it is difficult to draw a firm conclusion.

Although it is rare to fulfill all the Bradford Hill causal criteria, consistency, specificity, and analogy criteria exist in favor of a possible causal connection between TNF inhibitors and demyelination. Case reports documenting episodes of optic neuritis or demyelination following TNF inhibitor use began to emerge over a decade ago and continue to appear in the literature. Given the frequency and number of these reports, it is doubtful that new cases or small series would be reported in the neurology or ophthalmology literature presently. A trial of the TNF inhibitor lenercept in multiple sclerosis patients concluded that the TNF inhibitor–treated patients experienced earlier, more severe, and more frequent exacerbations compared with patients receiving placebo, providing analogous evidence of a connection between THF inhibitors and demyelination.

As with many aspects of our immunology understanding, the mechanism of demyelination with TNF inhibitors remains unknown despite possible hypotheses; this incomplete understanding does not influence the plausibility of a causal link.

In conclusion, the authors are to be commended for a fresh look at possible causality of a rare event; however, I would urge caution in disregarding a causal association between TNF inhibitors and episodes of demyelination at present.