A Self-Limited Diagnosis, and Whether to Shorten It
A biopsy has ruled out lymphoma and lupus and confirmed a benign, self-limited disease — but self-limited doesn't mean symptom-free, and how much steroid exposure is worth trading for a faster recovery has no settled answer.
Yuki T., a 23-year-old nursing student, presented with three weeks of tender cervical lymphadenopathy, fevers to 39.3°C most evenings, and fatigue severe enough that she has missed the past two weeks of clinical rotations. Her initial workup, driven by concern for lymphoma given the degree and persistence of her adenopathy, included an excisional lymph node biopsy that returned histiocytic necrotizing lymphadenitis without evidence of malignancy — Kikuchi-Fujimoto disease — after ANA and dsDNA antibody testing also came back negative, reasonably excluding a lupus-associated presentation, which shares real histologic and clinical overlap with KFD and needs to be considered in essentially every case.
The relief of ruling out lymphoma and lupus runs into a genuinely open question about what to do next, because “benign” and “self-limited” don't mean the same thing as “mild” for the weeks or months the disease actually takes to run its course. Kikuchi-Fujimoto disease reliably resolves on its own, most case series describing one to four months, and corticosteroids do not change that ultimate outcome. Yuki has no other significant medical history and had been, by her own description, entirely healthy before this started — the only recent change in her life is the stress of her final clinical rotations, which she and her attending both initially assumed explained the fatigue before the fevers and node swelling made that reading untenable. Her three weeks is the number worth reading carefully, because it is doing less work than it feels like it is doing. Against a course that typically runs one to four months, three weeks is not a disease failing to resolve; it is a disease a fifth of the way through its shortest expected span, and almost every patient who will eventually recover on her own looks exactly like this at week three. What steroids can do, per case-series data including a French cohort by Dumas et al., is shorten symptomatic duration in patients with significant fever burden and functional impairment — and this is where Yuki splits in two. On severity she sits squarely inside that cohort: 39.3°C most evenings and two full weeks of rotations lost is not a mild course by any reading. On duration she sits well outside it, since that cohort was assembled from protracted presentations and hers has not yet had the chance to become one. She meets half the description of the patients steroids helped, and the half she is missing is the half only time can supply.
Hematology follow-up, biopsy-confirmed diagnosis
This is going to resolve on its own — most series describe one to four months regardless of treatment, and corticosteroids don't change that ultimate course. She is three weeks in. At three weeks essentially every patient who is going to recover unaided still looks unwell, so 'she hasn't improved yet' isn't yet evidence of anything. I'd manage with NSAIDs and supportive care rather than exposing her to real steroid toxicity for a disease that isn't going to become anything more serious either way.
I don't disagree about the eventual outcome, but three weeks of nightly high fevers and missed clinical rotations is real, ongoing suffering, not just a waiting period to get through. If corticosteroids can meaningfully shorten that, I think withholding them purely because the disease is ultimately benign undersells how hard the actual weeks-to-months are to live through.
'Self-limited' is doing a lot of reassuring work in this conversation for someone who is, right now, missing her nursing rotations and running fevers most nights.
I don't think we need to pick one policy for every Kikuchi-Fujimoto patient today — the case-series literature itself, including Dumas et al.'s cohort, shows a real subset who need steroids and a larger subset who don't — and Yuki matches that cohort on severity while not yet matching it on duration, which is precisely the ambiguity we can't resolve from where we're standing today.
I'd start with NSAIDs now and set a defined two-week reassessment: if her fevers and functional impairment haven't meaningfully improved by then, add corticosteroids at that point. That reserves the real toxicity for a patient who's demonstrably not resolving on her own, rather than deciding for or against it today based on how she's doing right now, three weeks in, which may not predict where she is at week five.
Agreed: NSAIDs started today, with a firm two-week reassessment date set in advance for whether to add corticosteroids — the clinical pharmacologist's checkpoint approach let the hematologist's toxicity-avoidance concern and the primary care physician's present-suffering concern both stand without resolving which mattered more in the abstract.
NSAIDs continued alone; no corticosteroids added, consistent with the disease's expected self-limited course.
Corticosteroids added at that point — the primary care physician's concern for ongoing suffering becomes the operative consideration once the trial-of-NSAIDs data is actually in hand.