Secondary Logo

Journal Logo

Original Article

Metal Fume Fever: Characterization of Clinical and Plasma IL-6 Responses in Controlled Human Exposures to Zinc Oxide Fume at and Below the Threshold Limit Value

Fine, Jonathan M. MD; Gordon, Terry PhD; Chen, Lung Chi PhD; Kinney, Patrick PhD; Falcone, Gary; Beckett, William S. MD, MPH

Author Information
Journal of Occupational & Environmental Medicine: August 1997 - Volume 39 - Issue 8 - p 722-726
  • Free


Metal fume fever is one of a multitude of names for an acute, short-term respiratory and systemic syndrome induced by inhalation of metal fumes, chiefly zinc oxide fume.1,2 The syndrome is characterized by the appearance of symptoms a few hours after exposure to zinc oxide, the resolution of the illness within 24 hours, and the induction of clinical tolerance to successive exposures, which is generally lost after a few days' exposure hiatus. The predominant populations suffering from this disorder historically have been brass foundry and galvanized metal workers. Despite regulations limiting the exposure concentration to zinc oxide fume, an estimated minimum of 1500 to 2000 cases of metal fume fever occur each year in the United States, making it a common occupational illness.1-5

Although many of the cases of metal fume fever may be due to exposures to zinc oxide levels higher than the threshold limit value (TLV) and Occupational Safety and Health Administration's (OSHA's) permissible exposure limit (PEL) of 5 mg/m3 averaged over an 8-hour period, the lack of controlled exposure data underlying this exposure limit raises questions about the degree of protection afforded by the 5 mg/m3 exposure limit. Indeed, among the supporting data for the TLV are anecdotal reports of the occurrence of symptoms after exposure to 5 mg/m3.6 Yet aside from animal data and preliminary results in human subjects from our laboratory, there has been no controlled study testing whether exposure to zinc oxide at or below the current TLV induces adverse effects.7 One of the aims of this study, then, was to extend the investigation of the no- or limited-effects level of zinc oxide exposure.

The constellation of symptoms in metal fume fever fit well with a cytokine-driven process. Blanc, Kuschner and coworkers have hypothesized that such cytokines may be released from pulmonary macrophages stimulated by zinc oxide particles and have demonstrated increased levels of the pyrogenic cytokines interleukin (IL)-6 and tumor necrosis factor (TNF) in bronchoalveolar lavage of human subjects after welding.8,9 Given the systemic nature of metal fume fever, we hypothesize that IL-6 and TNF should be present in the circulation simultaneous with or preceding the onset of symptoms and fever. To observe clinical effects from exposure to zinc oxide at or below the 5 mg/m3 TLV and to track any rise in circulating IL-6 and TNF, we exposed volunteers to zinc oxide generated by a well-characterized furnace system.



The subjects were 13 healthy non-smoking volunteers (eight men and five women) with mean (±SD) age of 30.8 (±7.7) years. All subjects were informed of the risks of the experimental protocol and signed consent forms approved by the institutional review boards. None of the subjects had previously inhaled zinc oxide fumes or worked in professions in which they may have been exposed to zinc oxide fumes. Subjects were asked not to take nonsteroidal anti-inflammatory agents or acetaminophen for at least 24 hours prior to each exposure.


Subjects were exposed in a randomly ordered, single-blinded fashion to 0, 2.5, and 5 mg/m3 of zinc oxide. Each concentration was administered on a separate day, and each exposure was spaced by at least 48 hours (median interval time, 9 days) to minimize the induction of tolerance. Subjects inhaled zinc oxide while at rest for 2 hours through a snugly fitted mask (model #7910; Hans Rudolph, Inc., Kansas City, MO) vented to a hood. To standardize for diurnal variations, all exposures commenced between 8:30 and 9:30 AM.

The clinical effects of the exposures were assessed by having the subjects grade their symptoms on a visual analog scale reporting sheet immediately prior to each exposure and 3, 6, and 9 hours post-exposure. The symptoms listed were cough, fever, chills, flushing, fatigue, muscle aches, sweating, sickness, irritability, headache, dyspnea, stomach ache, and nausea. Itching was included as a negative control symptom because it has never been reported as part of the metal fume fever syndrome. The extent of symptoms were recorded by placing a slash mark through a 10-cm horizontal line. The headings of "slight," "mild," "moderate," and "severe" were centered at 2.5 cm, 4.5 cm, 7.5 cm, and 10 cm, respectively, along the line.

Subjects were instructed to record their oral temperatures with an electronic, digital read-out thermometer (Ivac Corp., San Diego, CA) prior to each exposure and a minimum of every 2 hours afterwards until they went to sleep that night. Each subject was assigned a single thermometer for use throughout the study. Because the circadian temperature rhythm is approximately 1.0°F, increases in temperature from the morning baseline that were greater than 1°F were regarded as fever.10

Venous blood samples were drawn before, 0, 3, and 6 hours after each exposure. The plasma portions were immediately separated and stored at -70°C for later analysis. Measurement of plasma IL-6 and TNF-α were performed using high-sensitivity enzyme-linked immunosorbent assay kits (R&D Systems, Minneapolis, MN) and a microtiter plate reader (Anthos 2000, Fallsburg, Austria).

Zinc Oxide Generation

Zinc oxide fume was generated by the method described by McCarthy and coworkers.11 In brief, zinc granules were heated to approximately 550°C in a furnace flowing with inert argon gas. The zinc vapors released were carried downstream to react with oxygen, yielding a supersaturated atmosphere of zinc oxide vapor, which condenses to ultrafine particles. These primary particles aggregate in chains to form secondary particles. Measurements performed with a differential mobility analyzer (model 3021; TSI, St. Paul, MN) have repeatedly demonstrated that these secondary particles have a 0.3-µm mass median diameter and 1.5 geometric standard deviation. The zinc oxide fume was diluted with metered room air filtered through a HEPA filter and an activated charcoal filter and humidified to 30% to 50% relative humidity using a cascade humidifier. Zinc oxide was collected during successive 20-minute intervals (six times total) from the manifold of the exposure system onto a Teflon™ filter (TX40HI20-WW; Pallflex Products Corp, Putnam, CT). Each filter was immediately weighed on a microbalance (Cahn C-30, Cerritos, CA) and, based on each filter weight, the furnace system was immediately adjusted to yield the desired output. The control exposure consisted of furnace gas composed of 97% filtered room air and 3% argon.

Statistical Analysis

For each outcome (temperature, symptoms, and cytokine level) and each exposure condition (0, 2.5, and 5 mg/m3), the difference between all post-exposure values and the pre-exposure value was computed. These data were then analyzed by two-factor analysis of variance (ANOVA) (subject x exposure) at each post-exposure time point to test whether mean responses differed across exposure conditions, controlling for between-subject differences. The ANOVA was followed by the post-hoc Dunnett???s test to determine which exposure conditions were significantly elevated compared with control. This parametric approach was justified by the fact that raw data for all outcomes were distributed in a reasonably symmetrical way within each analysis cell, and that there was little evidence for non-homogeneous variances. The primary symptom variable analyzed was the sum of all symptoms, excluding itching (the control symptom). In a post-hoc analysis, individual symptoms were tested for significance. We also tested for correlations between the maximal responses for the three outcomes: total symptom score, excluding itch; temperature; and IL-6 level. The digital thermometers used by the subjects to measure body temperature read-out in degrees Fahrenheit. We report changes in body temperature in degrees Fahrenheit because it remains the clinically relevant unit of measure in the United States.


Thirteen subjects entered the study, and 12 completed all three exposures. One subject dropped out, for unspecified reasons, after two exposures. Mean (±SD) time-weighted average zinc oxide concentrations for exposures designated as the 2.5 mg/m3 and 5 mg/m3 were 2.52 (±0.27) and 4.9 (±0.4), respectively (Figure 1).

Among the 12 subjects completing the protocol, all but two experienced a mild fever between 6 and 12 hours after exposure to zinc oxide fume (Figure 2). For nine subjects, the fever occurred after exposure to 2.5 mg/m3, and for seven subjects after exposure to 5.0 mg/m3. Three subjects mounted fevers after exposure to the control furnace gas, but each also recorded an even greater temperature rise after zinc oxide exposure (mean difference of 0.5°F). Four subjects recorded temperatures greater than 100°F. The peak temperature recorded was 100.9°F.

Fig. 1
Fig. 1:
Zinc oxide concentrations measured at 20-minute intervals throughout the 2-hour exposures. The target exposure concentrations were 2.5 mg/m3 and 5.0 mg/m3.
Fig. 2
Fig. 2:
Mean (±SE) changes in oral temperature from the pre-exposure baseline after inhalation exposures to air, 2.5 mg/m3, and 5 mg/m3 zinc oxide. * denotes significant difference (P < 0.05) from baseline.

Eleven of the 12 subjects reported symptoms after zinc oxide exposure in excess of those recorded after control exposures. With the 5 mg/m3 exposure, symptoms tended to peak in the early evening (Figure 3). The symptoms most consistently noted were fatigue, muscle ache, and cough. The mean change from the pre-exposure baseline for these symptoms scores at 9 hours after the 5 mg/m3 exposure were significantly different from those after the control exposure (Figure 4, A through C). Cough symptoms tended to have the earliest onset, followed by the systemic symptoms of fatigue and muscle ache. Symptom scores were mostly in the slight to mild range (approximately corresponding to values between 2 and 5), with only four subjects grading any symptoms as moderate (approximately corresponding to a value of 7) and none recording symptoms as severe (value of 10). Subjects reported all symptoms resolved by the next day after exposure. No subjects reported itching under any of the exposure circumstances.

Fig. 3
Fig. 3:
Mean (±SE) changes in total symptom score from pre-exposure baseline after inhalation exposures to air, 2.5 mg/m3, and 5 mg/m3 zinc oxide. Total symptom score represents the sum of 13 different symptoms scores. * denotes significant difference (P < 0.05) from baseline.
Fig. 4
Fig. 4:
Mean (±SE) changes in cough (A), fatigue (B), myalgia (C) scores from pre-exposure baseline after inhalational exposures to air, 2.5 mg/m3, and 5 mg/m3 zinc oxide. * denotes significant difference (P < 0.05) from baseline.

Comparison of the effects of exposure to 2.5 and 5 mg/m3 of zinc oxide fume showed that the mean (±SE) maximum rises in temperature from baseline were comparable between the two exposure concentrations: 1.35 ± 0.30°F for 5 mg/m3 and 1.23 ± 0.32°F for 2.5 mg/m3 zinc oxide. Symptom scores were significantly elevated only after the 5 mg/m3 zinc oxide exposure. No differences were noted between the responses of men and women.

Mean plasma IL-6 levels rose significantly from the pre-exposure baseline after both the 2.5 and 5.0 mg/m3 zinc oxide exposures, compared with the control exposure (Figure 5). The highest levels were recorded at 6 hours after zinc oxide exposure, preceding the fever peak. Although specimen availability limited the number of TNF determinations, no trends in TNF levels were apparent. Mean (±SE) TNF values (pg/mL) at pre-exposure and at 3 and 6 hours post-exposure were 0.92 (±0.35) (n = 6), 1.37 (±0.38) (n = 8), and 0.74 (±0.19) (n = 7), respectively, on the air day, and 1.3 (±0.17) (n = 10), 0.97 (±0.15) (n = 10), and 1.18 (±0.21) (n = 8) on the 5 mg/m3 zinc oxide day. Although mean IL-6 levels, temperature, and total symptom scores rose in parallel among subjects after exposure to 5.0 mg/m3 zinc oxide, the maximum increases in these values did not correlate with each other: IL-6 and total symptom score (excluding itch), r = 0.2, P = 0.55; temperature and IL-6, r = 0.38, P = 0.25; and total symptom score and temperature, r = 0.45, P = 0.15.

Fig. 5
Fig. 5:
Mean (±SE) changes in IL-6 (pg/mL) from pre-exposure baseline after inhalational exposures to air, 2.5 mg/m3, and 5 mg/m3 zinc oxide. * denotes significant difference (P < 0.05) from baseline.


The results of this study demonstrate that controlled inhalational exposure to zinc oxide for 2 hours at the current TLV of 5 mg/m3 produces fever and symptoms in human subjects. In addition, exposure to 2.5 mg/m3 zinc oxide for 2 hours also induces mild fever but not a significant increase in symptoms. The most consistently noted symptoms-fatigue, muscle ache, and cough-are part of the commonly catalogued constellation of symptoms of metal fume fever.12 Subjects varied widely in their responses, ranging from the development of moderately uncomfortable symptoms and fever to no reaction at all (one subject). Such intersubject variations in clinical response were noted by Drinker and coworkers in their classic laboratory exposures of human subjects to freshly generated zinc oxide.13

The induction of a clinical response after a 2-hour exposure to 5 mg/m3 zinc oxide generated by this furnace system is not surprising, given our earlier results with human subjects.7 Previous animal studies in this laboratory have demonstrated, moreover, that a 3-hour exposure to as little as 1 mg/m3 zinc oxide fumes can produce lung injury (unpublished data) and induce gene expression of the zinc-binding protein metallothionein in lung tissue.14 The significant rises in temperature and IL-6 levels, although not symptoms, after exposure to 2.5 mg/m3 suggest that a no-effect level of exposure may be considerably lower than the current standard of 5 mg/m3.

Consideration of the relevance of these controlled exposures to occupational exposures must take into account the physical properties of the fume as well as the exposure conditions. The submicronic size of the zinc oxide chain aggregates produced by our furnace system lies well within the size range of particles produced by welding (0.06 to 0.52 µm mass median aerodynamic diameter).15 The exposure in our laboratory was to a relatively constant concentration of zinc oxide inhaled continuously. This likely differs from many workplace patterns in which there are episodic high concentrations separated by prolonged periods of much lower concentration. Nonetheless, metal fume fever was induced after only one fourth the usual 8-hour workday duration and with a cumulative daily dose considerably lower than the OSHA PEL. The dose-response relationship observed is consistent with the opinion of the ACGIH TLV Committee (based on unattributed observations) that "zinc chills have occurred in foundry workers having inhaled concentrations below 5 mg/m3".6

The mechanism underlying the induction of metal fume fever by zinc oxide inhalation is unclear. Incubation of human peripheral blood monocytes with a zinc-rich elemental solution induces release of IL-6 but not TNF-α.16 Circulating zinc levels, however, do not rise above normal laboratory values after exposure to zinc oxide fumes,17 rendering it unlikely that inhaled zinc is absorbed and transported in amounts sufficient to cause a direct peripheral effect.

Alternatively, zinc oxide may exert its systemic effects indirectly through the lungs. Inhaled zinc oxide induces the release of cytokines in the lung compartment, as demonstrated in bronchoalveolar lavage fluid of subjects exposed to welding fumes.8 In order for the clinical syndrome of metal fume fever to occur, these mediators must enter the circulation. Although Blanc and colleagues did not detect consistent increases in circulating cytokine levels (IL-6, IL-8, TNF) after welding fume exposures,8,17 we observed significant increases in plasma IL-6 concentrations at 3 and 6 hours after exposure to freshly formed zinc oxide fumes. This difference in findings may be a result of our testing of more subjects at closer time points. The rise in plasma IL-6 from baseline levels to 6 hours post-exposure does fit with Blanc and coworkers' observation of higher levels in lavage fluid 8 hours vs 3 hours after zinc oxide exposure.

IL-6 is an important, multi-functional pulmonary cytokine that induces fever when injected intravenously or intrathecally in experimental animals18,19 and belongs to the family of glycoprotein 130 cell-surface receptor-activating pyrogens. In the lung it may be produced by macrophages, T or B lymphocytes, monocytes, fibroblasts, or mesangial cells, and its production is stimulated by IL-1 and TNF. Endotoxin, which may be responsible for the clinically similar "Monday morning fevers," induces IL-6 production by alveolar macrophages in vitro.20 Overlap between the effects of endotoxin and zinc is suggested by the finding that co-incubation of human peripheral blood monocytes with a zinc-rich trace element solution significantly augments endotoxin-induced IL-6 release.6

As more knowledge is gained of cytokine actions in the lungs, it appears likely that complex interactions of cytokines, rather than a simple chain reaction, participate in pulmonary/systemic syndromes such as metal fume fever. During the same time period when plasma IL-6 rose in these subjects, plasma TNF remained at pre-exposure constitutive levels. In experiments examining the time course of cytokine appearance in bronchoalveolar lavage fluid, the peals lavage TNF level occurred at 3 hours, well preceding the peak levels of IL-8 and IL-6.8 These findings raise the the possibility that inhaled zinc oxide stimulates the release of TNF, which in turn stimulates the release of IL-6 in the lungs. The IL-6 then enters the circulation, contributing to the production of the fever and symptoms of metal fume fever. Alternatively, it is possible that the IL-6 measured in the plasma was released from the liver or circulating cells. Current studies of bronchoalveolar lavage cytokines under identical exposure conditions in our laboratory may help to place these cytokines in temporal and anatomic sequence in the response to inhaled zinc oxide.


This study was supported by National Institute for Occupational Safety and Health Grant 5 RO1 OH-02987, a National Heart, Lung, and Blood Institute/DLD Preventive Pulmonary Academic Award (to Dr Beckett), Research Career Development Award ES 00256 (to Dr Gordon), National Institute of Environmental Health Sciences Center Grant ES00260, and The Polly Amenberg Levee Charitable Trust (to Dr Fine).


1. Gordon T, Fine JM. Metal fume fever. Occup Med. 1993;8:505-517.
2. Blanc P, Boushey HA. The lung in metal fume fever. Semin Respir Med. 1993;14:212-225.
3. Litovitz TL, Schmitz BF, Bailey KM. 1989 annual report of the American Association of Poison Control Centers National Data Collection System. Am J Emerg Med. 1990;8:394-422.
4. Litovitz TL, Bailey KM, Schmitz BF, et al. 1990 annual report of the American Association of Poisonn Control Centers National Data Collection System. Am J Emerg Med. 1991;9:461-509.
5. Litovitz TL, Holm KC, Bailey KM, Schmitz BF. 1991 annual report of the American Association of Poisonn Control Centers National Data Collection System. Am J Emerg Med. 1992;10:452-505.
6. American Conference of Governmental Industrial Hygienists. Documentation of the Threshold Limit Values and Biological Exposure Indices, 6th ed. Cincinnati, OH: ACGIH; 1991:1754-1757.
7. Gordon T, Chen LC, Fine JM, et al. Pulmonary effects of inhaled zinc oxide in human subjects, guinea pigs, rats, and rabbits. Am Ind Hyg Assoc J. 1992;53:503-509.
8. Blanc PD, Boushey HA, Wong H, Wintermeyer SF, Bernstein MS. Cytokines in metal fume fever. Am Rev Respir Dis. 1993;147:134-138.
9. Kuschner WG, D'Alessandro A, Wintermeyer SF, Wong H, Boushey HA, Blanc PD. Pulmonary responses to purified zinc oxide fume. J Investig Med. 1995;43:371-378.
10. Dinarello CA, Wolff SM. Pathogenesis of fever and the acute phase response. In: Mandell GL, Bennett JE, Dolin R, eds. Principles and Practices of Infectious Diseases. New York: Churchill Livingstone; 1995:530-536.
11. McCarthy JF, Yurek GJ, Elliott JF, Amdur MO. Generation and characterization of submicron aerosols of zinc oxide. Am Ind Hyg Assoc J. 1982;43:880-886.
12. Turner JA, Thompson LR. Health hazards of brass foundries. Public Health Bull. 1926:Number 157.
13. Drinker P, Thomson RM, Finn JL. Metal fume fever: IV. Threshold doses of zinc oxide, preventive measures, and the chronic effects of repeated exposures. J Ind Hyg. 1927;10:331-345.
14. Cosma G, Fulton H, DeFeo T, Gordon T. Rat lung metallothionein and heme oxygenase gene expression following ozone and zinc oxide exposure. Toxicol Appl Pharmacol. 1992;117:75-80.
15. International Agency for Research on Cancer. Chromium, nickel and welding. IARC Monogr Eval Carcinog Risks Hum. 1992; Volume 49.
16. Falus A, Beres J. A trace element preparation containing zinc increases the production of interleukin-6 in human monocytes and glial cells. Biol Trace Elem Res. 1996;51:293-301.
17. Blanc P, Wong H, Bernstein MS, Boushey HA. An experimental model of metal fume fever. Ann Intern Med. 1991;114:930-936.
18. Kishimoto T. The biology of interleukin-6. Blood. 1989;74:1-10.
19. Kluger MJ. Fever: role of pyrogens and cryogens. Pharmacol Rev. 1991;71:93-127.
20. Kotloff RM, Little J, Elias JA. Human alveolar macrophage and monocyte interleukin-6 production. Am J Respir Cell Mol Biol. 1990;3:497-505.
© Williams & Wilkins 1997. All Rights Reserved.