Cytokine Storm in a Phase 1 Trial of the Anti

Table 1. Table 1. Data for All Six Affected Patients on Transfer to the Intensive Care Unit (ICU).Table 2. Table 2. Median Results of Blood Tests for the Six Patients before Infusion and 8 and 16 Hours after Infusion of TGN1412.

All six patients who received the trial drug were male, with a median age of 29.5 years (range, 19 to 34) (Table 1). None had a notable medical history, and all were clinically well during the 2 weeks before the study; baseline laboratory values were normal (Table 2). Beginning at 8 a.m. on day 1, each volunteer received an intravenous infusion, 10 minutes apart, of either the study drug or placebo. Each infusion lasted 3 to 6 minutes. The six volunteers in the treatment group each received 0.1 mg of TGN1412 per kilogram of body weight, infused at a rate of 2 mg per minute; the remaining two volunteers received a similar volume of saline.

Initial Response after Infusion of TGN1412

Figure 1. Figure 1. Summary Timeline of the Main Events after Infusion of TGN1412.

The course is divided into four phases: cytokine storm, reactive, recovery, and steady state. ALT denotes alanine aminotransferase. Dashed lines represent the responses of Patients 5 and 6 (who were the most seriously ill).

A series of adverse effects began in the treatment group after infusion, starting with the onset of severe headache in five patients after a median of 60 minutes (range, 50 to 90), accompanied by lumbar myalgia in all six patients after a median of 77 minutes (range, 57 to 95) (Figure 1). Subsequently, during this early phase, the patients were restless and had varying degrees of nausea, vomiting, bowel urgency, or diarrhea (Table 1). Five subjects had short amnestic episodes associated with severe pyrexia, restlessness, or both. All patients had a systemic inflammatory response that included erythema and peripheral vasodilatation (the timing of which was undocumented), with recorded rigors in four patients at a median of 59 minutes (range, 58 to 120) after infusion. Hypotension (defined by a decline in systolic blood pressure of 20 mm Hg or more) developed in all patients a median of 240 minutes (range, 210 to 280) after infusion, accompanied by tachycardia, with maximal heart rates of 110 to 145 beats per minute. All patients received intravenous lactated Ringer's solution during this time. Body temperatures of 39.5 to 40.0°C were recorded a median of 280 minutes (range, 240 to 390) after infusion. At 300 minutes after infusion, Patient 1 had signs of respiratory failure, with tachypnea and a partial pressure of arterial oxygen (PaO₂) of 52 mm Hg while breathing ambient air; the PaO₂ increased with the addition of supplemental oxygen. Chest radiography revealed pulmonary infiltrates; these findings were not consistent with the expected response of a fit young man to the infusion of less than 4 liters of fluid at this stage. There was no clinical evidence of bronchospasm or laryngeal edema.

All patients were initially empirically treated in the independent clinical trials unit. A dose of 200 mg of hydrocortisone was administered intravenously in divided doses (with the initial 100-mg bolus a median of 331 minutes [range, 315 to 346] after infusion), in addition to 10 mg of chlorpheniramine intravenously, 1 g of acetaminophen intravenously, 4 to 8 mg of ondansetron intravenously, and 0.5 to 3.0 mg of metaraminol intravenously (in divided doses, titrated to effect). Blood samples were analyzed 8 hours after infusion at an off-site private laboratory (according to the study protocol) and therefore were not available as the situation evolved; the results were abnormal (Table 2).

Subsequent Events

After an initial recovery, Patient 6 became hypotensive (blood pressure, 65/40 mm Hg), and 12 hours after infusion, he had metabolic acidosis and marked respiratory distress with hypoxemia that was refractory to treatment with supplemental oxygen. He underwent intubation and mechanical ventilation, after which he was admitted to the intensive care unit (ICU) at Northwick Park and St. Mark's Hospital. He had severely abnormal hemodynamics, coagulation, and pulmonary function, with a PaO₂ of 84 mm Hg while breathing 100% oxygen (ratio of PaO₂ to the fraction of inspired oxygen, 84) (Table 1).⁵ Because there was concern that all patients would follow a similar course of rapid deterioration, all remaining patients were transferred to NHS ICU facilities 16 hours after infusion.

Further Treatment

Figure 2. Figure 2. Representative Chest Radiographs (Panels A and B) and Blood Smears (Panels C and D) of the Six Affected Patients.

All anteroposterior chest radiographs were similar in appearance, with interstitial infiltrates first noted 5 to 16 hours after infusion of TGN1412 (Panels A and B). The blood films did not show red-cell fragmentation but did show dysplastic changes in the neutrophils, including pseudo–Pelger–Huët anomaly (arrows, Panels C and D; May–Grünwald–Giemsa stain), which was first noted within 24 hours after TGN1412 infusion. Although toxic granulation, vacuoles, and Döhle's bodies were initially observed, later blood smears showed neutrophils that were hypogranular.

Between 16 and 20 hours after infusion of TGN1412, the patients had further signs of respiratory deterioration: all six had signs of tachypnea, use of accessory muscles, inability to complete spoken sentences, and bilateral pulmonary infiltrates on chest radiography (Figure 2A and 2B), and two had symptoms of dyspnea. There was also evidence of substantial renal impairment and disseminated intravascular coagulation, as indicated by an elevated prothrombin time, low fibrinogen level, high level of d-dimers, and decreased platelet counts in all six patients (Table 2). All patients had severe lymphopenia and monocytopenia, with sparing of neutrophils. Blood smears showed toxic granulation with Döhle's bodies and a dysplastic appearance of the neutrophils, with pseudo–Pelger–Huët anomaly (Figure 2C and 2D).

There was no clinical evidence of primary cardiogenic shock, nor was there bronchospasm, laryngeal edema, or cutaneous signs indicating anaphylaxis. There were no overt or focal neurologic symptoms or signs that suggested neurogenic vasodilatory shock. All electrocardiograms and echocardiograms were normal (Table 1), and there was no clinical indication for lumbar puncture or electroencephalography.

All patients received empirical treatment with 1 g of methylprednisolone sodium succinate intravenously a median of 16 hours (range, 15.5 to 17) after infusion with TGN1412, with subsequent doses 40 hours and 64 hours after. Because of the expected effects of TGN1412 on T cells, all patients were empirically treated daily for 3 days with an anti–interleukin-2 receptor antagonist antibody, daclizumab (Roche), beginning a median of 25.5 hours (range, 23.5 to 28.0) after infusion. This treatment was stopped after 3 days in the absence of TGN1412-induced lymphocytosis. In addition, potential activation of a histaminergic response was treated with 50 mg of intravenous ranitidine every 8 hours and 10 mg of intravenous chlorpheniramine maleate every 8 hours (continued from earlier doses).

Supportive Management

Patients 1 through 4 received continuous positive airway pressure of 10 cm H₂O by means of a tight-fitting face mask. Patients 5 and 6 underwent mechanical ventilation, with tidal volumes limited to 6 to 8 ml per kilogram of dry body weight and positive end-expiratory pressure maintained at 15 to 20 cm H₂O. All six patients had oliguria, metabolic acidosis, and increasing creatinine levels; they therefore received renal support by means of continuous venovenous hemodiafiltration with the use of a standard polyacrylonitrile membrane (Gambro Hospal U.K.) within 36 hours after their exposure to TGN1412. Dialysate rates were set to 1 liter per hour and were subsequently increased to 4 liters per hour.

All patients required the replacement of blood components by means of the infusion of fresh-frozen plasma and cryoprecipitate to correct coagulopathy. Owing to their severe lymphopenia, the patients were treated according to a protocol of infusions of irradiated red cells and platelets, as required, to prevent possible graft-versus-host disease.

Clinical Progression

Patients 1, 2, 3, and 4 continued to have intermittent fever, myalgia, and diffuse erythematous flushing for 48 hours, at which point their clinical symptoms and signs diminished markedly. Immunomodulatory treatment in these four patients was reduced to a tapering dose of intravenous hydrocortisone followed by oral prednisolone (total duration of corticosteroid treatment in each case, 21 days). Continuous venovenous hemodiafiltration was stopped after a median of 28 hours (range, 22 to 35), and continuous positive airway pressure was stopped after 4 hours in Patient 1 and after a median of 77 hours (range, 57 to 82) in Patients 2, 3, and 4 (Figure 4A and 4B through 7A and 7B in the Supplementary Appendix). Patient 2 was also successfully treated for presumed nosocomial Klebsiella pneumoniae bacteremia, isolated on day 6 after TGN1412 infusion.

Patients 5 and 6 had a more complex course, as detailed in the Supplementary Appendix. Although both patients initially had diminished erythema and fever 48 hours after infusion, they subsequently had recurrent fever, increased peripheral vascular permeability, and episodes of diffuse erythematous flushing lasting several days. Both patients required intubation and mechanical ventilation. Peripheral ischemia was observed in a glove-and-stocking distribution in Patient 6. It fluctuated over time, independently of the changing vasopressor dose. Most of the peripheral ischemia slowly resolved, except in patches of necrosis on the fingers of both hands and all the toes.

Over the next 30 days, all patients had generalized desquamation (most marked in Patients 5 and 6) and muscle weakness on discharge from the ICU. Five patients had late myalgia, headache after the discontinuation of corticosteroids, difficulties with concentration, and short-term difficulties in finding words (particularly names). Three patients had delayed hyperalgesia, and two had peripheral numbness. None had documented lymphadenopathy or splenomegaly while in the ICU or after discharge.

Hematologic and Immunologic Progression

Figure 3. Figure 3. Summary of Laboratory Results for the Six Patients during the First 30 Days (Panels A and B) and the First 5 Days (Panel C) after Infusion of TGN1412.

Panel A shows that C-reactive protein and serum creatinine levels increased rapidly during the first 48 hours after infusion, with a concomitant decline in the platelet count starting within the first 8 hours and persisting for at least 5 days. Alanine aminotransferase levels increased slowly, starting within the first 48 hours, and peaked between 10 and 25 days after infusion, when the patients had recovered from the acute illness. Panel B shows that levels of CD3+, CD4+, and CD8+ T-cell subgroups were undetectable within the first 24 hours after infusion, followed by a first peak at day 5 and a second peak at day 15, with a leveling off to near-preinfusion levels by day 30. Monocyte numbers also fell in the short term but increased to above the normal range 10 to 16 days after infusion. Neutrophil counts were relatively constant immediately after infusion and then increased, as expected, with increasing stress and corticosteroid use. Panel C shows that, during the first 4 hours after infusion, the first cytokine to increase substantially was TNF-α (2.8 pg per milliliter at 0 hour, 1760.1 at 1 hour, and 4675.9 at 4 hours), followed by interferon-γ (7.1 pg per milliliter at 0 hour, 43.9 at 1 hour, and ≥5000 at 4 hours) and interleukin-10, 8, 6, 4, 2, 1β, and 12p70. All data are medians. I bars represent interquartile ranges. Dashed lines represent the upper limit of the normal reference range (where only one dashed line is shown) or both the upper and lower limits. Time points with single values were excluded. To convert values for creatinine to milligrams per deciliter, divide by 88.4.

The laboratory values for the six patients are summarized in Figure 3; data on the clinical course of each patient are provided in the Supplementary Appendix. Severe thrombocytopenia was observed, initially accompanying disseminated intravascular coagulation but persisting even after the other clotting values normalized (Figure 3A, and Figure 4A and 4B through 9A and 9B in the Supplementary Appendix). All patients had mild normocytic anemia that persisted beyond discharge from the ICU, followed by a slow recovery. Neutrophil numbers initially were preserved and then increased in response to corticosteroids (Figure 3B), but they were dysplastic in appearance (Figure 2C and 2D), a feature that eventually resolved. By contrast, marked lymphopenia and monocytopenia were noted in all patients 8 hours after TGN1412 infusion (Table 2).

Lymphocyte numbers were too low to allow for the measurement of cell subgroups 1 day after infusion. Subsequent blood tests showed increasing levels of CD4+ and CD8+ T cells (Figure 3 through 9 in the Supplementary Appendix), CD19+ B cells, and CD16+ presumed natural killer cells, starting 48 hours after infusion. In Patients 1, 2, 3, and 4, who recovered the most rapidly, T-cell recovery occurred in a CD4+:CD8+ ratio of 1:1, with a temporary rise to levels just above normal in two patients (Figure 4D and 7D in the Supplementary Appendix). Patients 5 and 6, the two who were most severely ill, had a slower recovery, with lower overall numbers of T cells (Figure 8 and Figure 9 in the Supplementary Appendix) and a CD4+:CD8+ ratio of 2:1.

The lymphocyte and monocyte nadirs in each patient occurred within 24 hours after TGN1412 infusion, overlapping with the cytokine storm (Figure 3B, and Figure 4C through 9C in the Supplementary Appendix). A dramatic increase in the level of tumor necrosis factor α (TNF-α) was observed in all patients within an hour after TGN1412 infusion, followed by elevations in the level of interleukin-2, 6, and 10 and interferon-γ within the first 4 hours after infusion (Figure 3C, and Figure 4C through 9C and 10 in the Supplementary Appendix). This cytokine release resolved after the first doses of hydrocortisone and methylprednisolone, and in Patients 1, 2, 3, and 4 the values normalized within 2 days. By contrast, in Patients 5 and 6, the cytokine storm was prolonged by 1 to 2 days; discrete elevations in the interleukin-6 and interleukin-4 levels, out of proportion to those noted in the other patients, were observed.