A companion diagnostic doesn't clear on its own clock

A diagnostic test that happens to be useful alongside a particular drug isn't automatically a companion diagnostic in FDA's sense of the term, and the difference isn't labeling flourish. FDA's regulations define an in vitro companion diagnostic device, at 21 CFR 809.3, as an in vitro diagnostic device that provides information essential for the safe and effective use of a corresponding therapeutic product — a definition built around necessity, not usefulness. A test a clinician could reasonably skip and still prescribe safely doesn't meet it, however informative that test might be. Once a diagnostic does meet the definition, FDA's guidance describes an expectation that follows directly from it: the diagnostic and the therapeutic product it's essential to are supposed to reach approval together, not as two submissions running on independent schedules that happen to land in the same review division.

The word doing the work is “essential”

FDA's final guidance on in vitro companion diagnostic devices, issued August 6, 2014, built the 21 CFR 809.3 definition around a single operative word. The therapeutic product's safe or effective use has to actually depend on the diagnostic's result — selecting patients likely to benefit, ruling out patients at risk of a serious adverse reaction, or monitoring response to adjust a dose. A biomarker test that correlates with outcomes, or that a clinician might reasonably consult, doesn't clear that bar by itself; the question the definition asks is whether the therapeutic can be used safely and effectively without the test's result, not whether the result is helpful. Sponsors who label a test a companion diagnostic before establishing that dependency are asking FDA to accept a marketing description the definition doesn't yet support.

Contemporaneous approval is the expectation, not an absolute

Once a diagnostic clears that bar, FDA's guidance states a preference explicitly: in most circumstances, an IVD companion diagnostic and its corresponding therapeutic product should be approved or cleared contemporaneously, for the use reflected in the therapeutic's labeling. That's not a formality about paperwork arriving on the same day so much as a substantive judgment — if the therapeutic's safe use actually depends on the diagnostic, approving the drug first leaves prescribers making that dependent decision without the tool the label says they need. FDA's guidance does carve out narrower circumstances, generally involving a serious or life-threatening condition with an unmet medical need, where a therapeutic can proceed ahead of its diagnostic under specific conditions. The guidance treats that as a deliberate exception built for a specific situation, not a routine option a sponsor reaches for out of scheduling convenience.

The labeling tie runs both directions

The dependency isn't a one-way reference. The therapeutic product's own labeling, under 21 CFR 201.57, and the IVD's labeling, under 21 CFR 809.10, are each supposed to reflect the other — the drug label states that use of the diagnostic is required, and the diagnostic's label states what it's cleared or approved to support. A mismatch between the two, drafted separately by two sponsors or two review teams working from different assumptions about timing, is the kind of gap that surfaces late precisely because each label reads as complete on its own until it's checked against the other.

Where codevelopment timing actually breaks

FDA's separate guidance on codevelopment principles exists because this coordination is genuinely hard to sequence well. A diagnostic's analytical and clinical validation generally has to be locked before the therapeutic's pivotal trial can rely on it to select patients — which means the diagnostic can't be an afterthought built after pivotal data already exists without retrofitting validation onto a device that was already used, under a research protocol, to generate the trial's own results. That retrofit is a materially harder position than a therapeutic moving quickly through an expedited pathway ever anticipates on the diagnostic side, and it's usually the diagnostic, reviewed as its own PMA or de novo file, that ends up setting the pace for both.

Where this meets the file

A codevelopment tracker built around this pathway needs the essential-use determination recorded as its own decision, distinct from a general note that the diagnostic is “relevant” to the drug, plus a field confirming the two labels actually cross-reference each other rather than assuming they will by the time both are final. A companion diagnostic codevelopment worksheet built around FDA's own essential-use and contemporaneous-approval language is previewed in the launch catalog. If your program tracks this differently, the shelf takes that correction directly.