Why Lactate Infusions Trigger Panic but Hard Exercise Rarely Does

Lactate has haunted psychiatry for decades. Infused sodium lactate can reliably provoke full panic attacks in susceptible patients, yet the same molecule surges to far higher concentrations during hard exercise with almost no panic at all. A new mini review in Frontiers in Psychiatry argues that this long-standing paradox has been framed incorrectly, and that neither the origin of the lactate nor its concentration in the blood can explain why the molecule panics some people and calmly fuels others.

The review, authored by Xinze Li, Shikun Wang and Weifu Chen and published on 29 October 2026, examines three physiological situations side by side: endogenous lactate produced during exercise, exogenous lactate delivered rapidly in psychiatric challenge tests, and exogenous lactate clamped at a steady level during rest or exercise. The authors’ central point is deceptively simple. If lactate itself were a sufficient panicogen, the clamp studies should have triggered attacks. They did not.

The numbers make the puzzle vivid. In a pivotal study of patients with panic disorder, supramaximal exercise drove mean blood lactate to approximately 10.7 millimoles per liter, yet only one of 24 participants experienced a panic attack. Historical sodium lactate challenges, by contrast, reported much higher panic rates, and the exercise study explicitly noted that its lactate levels exceeded those usually achieved during infusion. Meanwhile, in healthy volunteers, acute exercise-induced hyperlactatemia did not potentiate a subsequent 35 percent carbon dioxide challenge, and acute aerobic exercise actually reduced experimentally induced panic, attenuated carbon dioxide-evoked panic in patients with panic disorder, and reduced cholecystokinin-4-induced panic in both patients and controls. These studies are small, the authors concede, but their direction of effect is difficult to reconcile with lactate acting alone as a panic trigger.

The clamp experiments sharpen the argument. In one protocol, researchers clamped lactate at roughly 4 millimoles per liter during 90 minutes of rest followed by 90 minutes of exercise in six healthy active men, deliberately separating lactate availability from workload and demonstrating increased lactate oxidation and glucose sparing. A related nine-man training study used resting targets of 3.5 to 4.0 millimoles per liter with exercise-matched concentrations, and perceived exertion did not rise with the clamp. A further study in six healthy men documented approximately 4 millimoles per liter at rest and 4 to 7 during exercise, along with volume shifts and mild systemic alkalosis, again without reported panic. A contemporary resting study in eight healthy men used racemic sodium lactate for six hours, reaching 2.7 plus or minus 0.5 millimoles per liter postabsorptively, also without reported panic. Crucially, the review stresses that none of these metabolic studies used a dedicated panic diagnostic protocol, so absence of reported panic is not a prospectively ascertained zero-attack rate, and tolerance in healthy men cannot establish tolerance in panic disorder.

If lactate concentration and exogenous delivery are ruled out as universal explanations, what remains? The review systematically dismantles several simple candidates. Infusion site cannot explain the contrast, since clamps and psychiatric challenges both use peripheral arm veins and no randomized comparison of infusion sites exists. Mild alkalosis is insufficient, because the clamp studies produced alkalosis without panic. Fasting state does not rescue the story either, since the clamp studies used standardized meals and postabsorptive conditions rather than fed participants who were protected by having eaten. Blood lactate elevation, exogenous administration, fasting, mild alkalosis and absence of exercise are each, the authors conclude, insufficient as universal explanations.

Formulation, however, has real experimental support. In a randomized human study, sodium lactate and 3 percent sodium chloride produced similar panic incidence and similar sodium increases, despite alkalosis with lactate and mild acidosis with chloride, suggesting that under those conditions lactate itself was unnecessary, although sodium and osmolality were never independently isolated. More strikingly, a randomized double-blind crossover study in 20 patients and 14 controls found less panic when lactate was delivered in 5 percent dextrose than in saline, and no controls panicked at all. That finding provides direct evidence that the vehicle accompanying lactate can modify challenge sensitivity. Stereoisomer composition adds another layer: the tolerated clamps used L-lactate formulations, whereas historical psychiatric challenges frequently used racemic D,L-lactate, and in a 19-patient pilot, pure D-lactate induced panic in roughly half of participants, with hypocapnia and alkalosis. The tolerated 2024 metabolic infusion was also racemic, so no single isomer story holds. The authors also note that whether unneutralized lactic acid provokes human panic as effectively as equimolar sodium lactate remains unknown, with no direct comparison identified in panic disorder.

Dose and timing fare no better as explanations. The standard challenge delivers a 0.5 mole per liter solution at 10 milliliters per kilogram over 20 minutes, equivalent to 5 millimoles per kilogram of total lactate if completed, but the review emphasizes that this describes a protocol, not a required blood threshold or latency. No universal blood lactate threshold or minimum infusion duration for panic has ever been established. The clamps produced overlapping concentrations for much longer durations without reported panic, high exercise lactate seldom provoked attacks, and even patients who did not panic during challenges still developed substantial lactate, pyruvate and pH changes. Concentration trajectory, infusion rate, cumulative sodium load and time to attack, the authors argue, require joint analysis, particularly because early stopping at panic creates outcome-dependent exposure. Biological sex likewise fails as a stand-alone explanation: in a 214-person challenge sample, sex added no predictive information, although baseline blood gas measures differed between women and men and few female controls limited the inference.

What the authors propose instead is an interoceptive context-gating model. Interoception encompasses the sensing, integration and interpretation of signals from within the body, with visceral signals represented along the insular cortex and the anterior insula integrating bodily state with attention, emotion, context and action. In anxiety, the core disturbance may be less an inaccurate reading of the body than the excessive precision assigned to bodily prediction errors that could signal threat. Applied to lactate, the model holds that physiological perturbation and host vulnerability interact with expectations and perceived control. During exercise, motor commands, proprioception and pacing may make bodily arousal predictable and therefore non-threatening. During a well-explained resting clamp, procedural expectations and a sense of control could similarly reduce threat without any motor prediction involved. Conversely, panic-prone participants may interpret challenge-associated breathlessness or dizziness catastrophically. The authors are careful to label these as testable contextual moderators, not mechanisms already demonstrated, and they note that tolerated resting infusions rule out motor prediction as a necessary condition for avoiding panic.

The insular component of the hypothesis remains explicitly unproven. Anterior-insula lesions impair respiratory interoceptive attention, and cardiac attention and anxiety recruit insular regions, which makes insular participation plausible, but no cited study demonstrates differential insular gating across all three lactate conditions. The review insists that any gating account must be tested against chemical and respiratory explanations in matched-exposure studies, and it flags proton spectroscopy’s inability to distinguish circulating from locally produced lactate as a further limitation. Animal findings, including amygdalar ASIC1A contributions to carbon dioxide-evoked fear in mice and HCAR1-mediated effects on neuronal activity in rodent and human tissue, concern different timescales and endpoints and do not establish human insular causality.

The review also maps the boundaries of the question. Sepsis, where endogenous lactate rises during life-threatening illness, remains a separate case: no prospective study has paired acute panic assessment with serial lactate measurements in septic patients with pre-existing panic disorder, and delirium, sedation and overlapping respiratory disturbances make assessment treacherous. Pyruvate, measured alongside lactate in old challenge studies, has never been tested in a controlled human panic comparison against lactate alone. The authors sketch what a decisive study would look like: a within-participant comparison of exercise alone, an L-lactate clamp at rest, and the same clamp during exercise, in patients with panic disorder and matched controls, with matched concentration trajectories, characterized formulation and nutrition, prespecified validated panic outcomes, arterial or arterialized sampling, separate D and L assays, measured osmolality, blood gases and ventilation, and expectations and perceived control measured rather than inferred. Their bottom line is measured: lactate exposure is neither uniformly panicogenic nor uniformly protective, existing controls exclude several simple universal explanations while supporting contributions from formulation and host state, and conditional context gating stands as a falsifiable hypothesis rather than a completed resolution of one of psychiatry’s most enduring metabolic puzzles.

Subject of Research: Why sodium lactate provokes panic in susceptible patients but exercise-induced hyperlactatemia seldom does, and an interoceptive context-gating hypothesis to explain the contrast

Article Title: Lactate and panic across infusion and exercise contexts: an interoceptive context-gating hypothesis

Article References: Li, X., Wang, S., & Chen, W. (2026). Lactate and panic across infusion and exercise contexts: an interoceptive context-gating hypothesis. Frontiers in Psychiatry, 17, 1975318. https://doi.org/10.3389/fpsyt.2026.1975318

Image Credits: AI Generated

DOI: 10.3389/fpsyt.2026.1975318

Keywords: lactate, panic disorder, sodium lactate challenge, exercise, lactate clamp, interoception, insular cortex, predictive coding, formulation effects, acid-base regulation, anxiety, panic