research March 14, 2026 7 min read

The Smoke Alarm That Won't Stop: How IFN-γ Reveals Long COVID's Hidden Fire

long-covid biomarkers immunology IFN-gamma treatment-pipeline
The Smoke Alarm That Won't Stop: How IFN-γ Reveals Long COVID's Hidden Fire

A Fire Alarm Stuck On

When you fight off a virus, your immune system sounds the alarm. Interferon-gamma — IFN-γ — is one of the loudest signals: a cytokine released by T cells that coordinates the attack, kills infected cells, and activates macrophages. When the virus clears, the alarm stops. IFN-γ returns to baseline. The war is over.

In Long COVID patients, the alarm never stops.

A University of Cambridge study published in Science Advances found that Long COVID patients produce IFN-γ spontaneously — without any external provocation — for up to 31 months after infection. Their immune systems behave as though the war is still raging, even when no virus can be detected.

This isn't just another elevated cytokine. Three independent research groups, using different methods in different tissues, have now converged on IFN-γ as a central driver of Long COVID pathology. Together, they tell a story that runs from blood to lungs to the organelles inside individual cells.

The Cambridge Discovery: A Signal in the Blood

Dr. Benjamin Krishna's team at Cambridge used highly sensitive FluoroSpot assays — more precise than standard ELISA tests — to measure cytokine production from immune cells of 111 COVID-confirmed patients. Fifty-five had Long COVID with severe symptoms persisting at least five months.

What they found was striking. Long COVID patients' peripheral blood mononuclear cells (PBMCs) released IFN-γ spontaneously. No viral peptides were added. No stimulation was needed. The cells just kept producing this inflammatory signal, month after month, for up to 31 months.

The cellular mechanism was specific: CD8+ T cells were producing the IFN-γ, and they required antigen presentation by CD14+ monocytes via MHC class I. This isn't random inflammation — it's a targeted immune response. The T cells are behaving as though they're still seeing viral antigen, presented by monocytes that may be harboring viral remnants.

Critically, when patients improved, their IFN-γ levels dropped. More than 60% of patients whose symptoms resolved showed corresponding IFN-γ decline to baseline. The fire alarm turned off when the fire went out.

The Symptom Mirror

Here's a detail that makes this finding visceral rather than abstract.

IFN-γ has been used therapeutically to treat hepatitis C and certain cancers. The side effects of IFN-γ therapy are well documented: fatigue, fever, headache, muscle pain, depression, cognitive difficulty. Read that list again. It is, almost exactly, the symptom profile of Long COVID.

As Dr. Krishna noted: "These symptoms are all too familiar to Long COVID patients."

This isn't proof that IFN-γ causes Long COVID symptoms — correlation is not causation. But it's powerful circumstantial evidence. We know that giving healthy people IFN-γ produces Long COVID-like symptoms. We know that Long COVID patients have persistently elevated IFN-γ. And we know that when IFN-γ drops, symptoms resolve.

The Lung Confirmation: IFN-γ as Driver, Not Just Marker

If the Cambridge study showed IFN-γ as a signal in the blood, a second study in Science Translational Medicine showed it as a weapon in the lungs.

Researchers used single-cell RNA sequencing on bronchoalveolar lavage samples from patients with respiratory Long COVID (PASC). They found a distinct pro-fibrotic monocyte-derived macrophage response — macrophages that were actively driving scar tissue formation in the lungs. The interactions between these pulmonary macrophages and resident T cells were abnormal, and IFN-γ emerged as the key mediating signal.

Then came the critical experiment. In a mouse model of respiratory PASC, the team administered anti-IFN-γ treatment after acute infection had resolved. The result: reduced lung inflammation and reduced fibrosis.

This moved IFN-γ from biomarker to therapeutic target. It's not just a marker of something going wrong — it's a driver of the damage itself.

The Peroxisome Connection: Where IFN-γ Meets Cellular Destruction

A third paper, published in Science by Xiaoqin Wei and colleagues at the University of Virginia, added the deepest mechanistic layer yet.

The UVA team discovered that excessive interferon signaling — the very signal IFN-γ amplifies — remodels and destroys peroxisomes inside alveolar macrophages. Peroxisomes are small organelles critical for lipid metabolism and detoxifying reactive oxygen species. Without functioning peroxisomes, the macrophages can't resolve inflammation. Instead, they activate the inflammasome, opening Gasdermin D pores that pour out IL-1β — one of the most potent inflammatory cytokines.

The downstream consequences cascade: IL-1β drives accumulation of KRT8-high dysplastic epithelial progenitor cells. These dysfunctional progenitors can't properly regenerate lung tissue. Instead, they drive chronic inflammation and fibrotic remodeling — the same pathology seen in Long COVID patients with persistent respiratory symptoms.

Human lung samples from PASC patients confirmed that peroxisome impairment persists long after acute infection. And in mice, pharmacological enhancement of peroxisome biogenesis using sodium phenylbutyrate (4-PBA) — an FDA-approved drug for hyperammonemia — restored macrophage function, reduced inflammation, and mitigated fibrosis.

Three Papers, One Story

Step back and see what these three studies show together:

Study Tissue Finding Journal
Krishna et al. (Cambridge) Blood (PBMCs) Persistent spontaneous IFN-γ release; correlates with symptoms Science Advances
Hsieh et al. Lungs (BAL) IFN-γ drives pro-fibrotic macrophages; anti-IFN-γ rescues mice Sci Transl Med
Wei et al. (UVA) Lungs (macrophages) IFN signaling destroys peroxisomes → inflammasome → fibrosis Science

IFN-γ is the signal in blood that marks the disease. It's the driver in lung tissue that perpetuates fibrosis. And it's the upstream trigger that destroys the very organelles that would otherwise resolve the inflammation.

Three independent groups. Three different methodologies. Three levels of biology — systemic, tissue, and subcellular. All pointing to the same molecule.

The Treatment Connection: JAK Inhibitors Are Already in Trials

IFN-γ signals through the JAK-STAT pathway — specifically JAK1 and JAK2. This is critical because JAK inhibitors already exist, are already FDA-approved for other conditions, and are already being tested in Long COVID trials:

REVERSE-LC (baricitinib, JAK1/2 inhibitor): The largest JAK inhibitor trial for Long COVID. Expanded in February 2026 from 4 to 17 enrolling sites, now led by Vanderbilt with Stanford support. 550 patients. Neurocognitive data expected November 2026, full results July 2027. New sites include Brigham & Women's, NYU Langone, UCLA, and Yale.

LC-REVITALIZE (upadacitinib + pirfenidone): 348 patients across 6 countries. Combines a JAK1 inhibitor with an anti-fibrotic — a combination that directly addresses the IFN-γ → peroxisome → fibrosis chain described by Wei et al.

CLEAR-LC (abrocitinib, JAK1 inhibitor): Beth Israel Deaconess Medical Center.

These trials were designed before the three-paper convergence was visible. They were based on broader immunomodulation rationale. But the IFN-γ evidence now provides specific mechanistic justification: JAK inhibitors should suppress IFN-γ signaling, which should reduce the persistent immune activation that drives symptoms, macrophage dysfunction, and fibrosis.

Additionally, the Wei et al. finding opens a second therapeutic avenue: sodium phenylbutyrate to restore peroxisome function downstream of IFN-γ damage. This drug is already FDA-approved, cheap, and orally available. It addresses the consequences of IFN-γ signaling rather than the signal itself — a complementary approach.

What We Don't Know

Honesty requires noting the gaps:

Why does IFN-γ persist? The leading hypothesis is viral persistence — SARS-CoV-2 or its antigens hiding in tissue reservoirs, continuously stimulating T cells. The Cambridge finding that monocytes are presenting antigen via MHC class I supports this. But direct evidence of persistent replicating virus in most Long COVID patients remains elusive.

Is it IFN-γ alone? Unlikely. Long COVID involves multiple overlapping mechanisms — autoantibodies, microbiome disruption, vascular dysfunction, mast cell activation. IFN-γ may be the dominant driver in some subtypes but not others. The eight trajectory profiles we described yesterday suggest that different patients may have different primary drivers.

Will suppressing IFN-γ help — or backfire? IFN-γ is essential for fighting infections and tumors. Long-term JAK inhibitor use carries real risks: increased infection susceptibility, thrombosis, and malignancy. The trials will need to show that targeted IFN-γ suppression in Long COVID patients provides net benefit.

What Comes Next

For the first time in Long COVID research, we have a molecule that shows up in three independent studies as both a diagnostic marker and a mechanistic driver, with drugs already in trials that target its signaling pathway.

This doesn't mean IFN-γ is the whole story. But it means we've moved from "we don't understand Long COVID" to "we have a specific, testable, druggable hypothesis." That's how medicine advances — not through silver bullets, but through the slow convergence of independent evidence pointing in the same direction.

The smoke alarm is still going off. Now at least we know which alarm it is — and we have tools to reach it.

Sources: