Patch Testing With a Large Series of Metal Allergens

Findings From More Than 1,000 Patients in One Decade at Mayo Clinic

Mark D.P. Davis; Michael Z. Wang; James A. Yiannias; James H. Keeling; Suzanne M. Connolly; Donna M. Richardson; Sara A. Farmer


Dermatitis. 2011;22(05):256-271. 

In This Article


We report our experience with more than 1,000 patients who underwent patch testing with metals during the past decade at Mayo Clinic. Many metals with high reaction rates (eg, gold, nickel, and cobalt) were familiar causes of allergic patch-test reactions. However, other metals with high reaction rates—such as manganese chloride 2% (allergic rate, 20.2%), palladium chloride (13.3%), and Ticonium (10.8%)—were less familiar. Some metal preparations notably yielded no positive reactions (titanium, Vitallium, and aluminum powder), and others had extremely low rates of reaction (zinc [allergic rate, 0.1%], ferric chloride [0.1%], and tin [0.3%]).

Selection of Patients for Patch Testing

All patients who were patch-tested with the metal series were suspected of having an allergy to metals. Because these patients represented a select population, it was not surprising that rates of allergic patch-test reactions to metals such as nickel and gold sodium thiosulfate were higher in this population than in the general population being patch-tested.[3,12]

Patch testing with metals was performed for divergent reasons and in many settings. Indications for patch testing included oral disease, generalized dermatitis, and dermatitis affecting almost any part of the body. For example, one reason for patch testing was to determine whether implanted metal devices were associated with positive patch-test reactions.[5] Our experience in patch-testing patients with oral disease is described in previous published reports. Yiannias and colleagues[14] and Torgerson and colleagues[15] proposed that contact sensitivities are important in patients with the diagnosis of oral lichen planus; the removal of metals may improve the condition. Dental alloys are known to release sensitizers in saliva;[16–20] such haptens may induce allergic reactions in the vicinity of the mouth.

Extended Patch-test Readings

A particular strength of our study was that late readings (up to 2 weeks after the application of metal allergens) were required throughout the study period at Mayo Clinic in Rochester (although not at Mayo Clinic in Arizona or in Florida, where patients are advised to call if they have a later reaction). We have previously reported our data regarding late patch test readings.[21] In our experience, patch-test results were negative at day = but became positive by day 7 or later for the following metals: gold sodium thiosulfate 0.5%, palladium chloride 2%, potassium dicyanoaurate 0.1% aqueous, gold sodium thiosulfate 0.25%, beryllium sulfate tetrahydrate 1% aqueous, nickel sulfate hexahydrate 2.5%, cobalt sulfate 1%, and gold chloride 0.5% in alcohol. These late readings increased our detection rate for allergic patch-test reactions to metals.


Which allergic patch-test reactions have any relevance? The interpretation of patch-test reactions continues to be the biggest obstacle when patch-testing with metals, especially in the settings described. Should a patient with oral lichen planus adjacent to gold crowns and with a positive allergic patch-test reaction to gold have the crowns removed? Should a patient with a loosened titanium hip joint and a positive allergic patch-test reaction to titanium or cobalt chromium have the joint removed? The only way to accurately judge relevance would be to determine whether improvement occurred after removal of the suspected allergen; although efforts are made to estimate relevance, it is difficult to ascertain. True relevance values can be assigned only if patients avoid the identified allergen and subsequently report the result of avoidance.[5]

In the present study, physicians reviewed the results of the readings and discussed the relevance of their patch-test results, given the exposures. Relevance was designated by a code indicating that the physician thought the reaction had definite relevance, questionable (probable or possible) relevance, or no relevance. Although such designations were made in this study (see Table 3), they can be regarded with some skepticism because it is extremely difficult to assign true relevance at patch testing. These values are inherently subjective and indicate the physician's opinion at the time of the patch-test readings.

Metals of Different Salts and Concentrations

Given that we tested metal preparations of different salts and in different concentrations, can this wealth of data be used to determine which preparation for each metal was the most valuable for the diagnosis of allergic contact dermatitis? The answer is complicated and unclear.

Figure 1 and Table = show the data for different concentrations and different salts; allergic patch-test reaction rates varied considerably for different concentrations and salts. In general, the greater the concentration, the higher the allergic patch-test reaction rate. We also observed intriguing heterogeneous results within individuals; that is, patients frequently reacted to one salt or concentration of a metal but not to another. For example, four patients reacted to gold sodium thiosulfate at the 0.25% concentration but not to that same metal at the 0.5% concentration (see Table 5). The reason for this variation in response is unclear.

Figure 1.

Comparison of numbers of patients with positive patch-test reactions to different concentrations and salts of tested allergens. (alc = in alcohol; aq = aqueous.)

Vitallium yielded no positive reactions, yet the alloy consists mostly of cobalt and chromium, two metals that show relatively high reactivity when tested alone. There are many possible explanations for this (such as the low release of metals from this alloy, or the concentrations or salts that were used).

Sex Differences and Metal Allergic Patch-test Reaction Rates

Of the metals tested, nickel, gold, palladium, and cobalt produced allergies more commonly in females. This finding was consistent with previously published data concerning nickel,22–24 gold,24–26 palladium, and cobalt.[22,27] These metals all are commonly used in jewelry; for example, palladium (one of the platinum group metals) is used to lighten the color of white gold, and cobalt is often used to produce blue colors in jewelryrelated materials such as enamels.

Also consistent with previous reports, rates of allergic patch-test reactions to mercury compounds were not statistically different between the sexes.[22] In contrast, allergic patch-test reactions to chromium were significantly more common in males (p = .048). This finding was consistent with previously published data[28–30] and is presumed to be due to occupational exposures.[13,31] Chromium is found in materials used in male-prevalent occupations (eg, cement-production materials, paints, varnishes, and leather-tanning materials).

Age Differences and Metal Allergic Patch-test Reaction Rates

The frequency of allergic patch-test reactions to certain metals (eg, nickel, palladium, and cobalt) was highest among younger individuals and diminished with age. This finding was concordant with the results of previous reports.[22,28] We speculate that exposure to these metals (eg, through jewelry) is perhaps more common among younger persons. Of interest, after regulations on nickel-releasing jewelry were established in Europe, the prevalence of nickel allergy in younger age groups decreased.[32,33] Detailed differences in patch-test reactions by age are not shown here owing to space constraints.

Strengths and Limitations of the Study

We present data on a large number of patients patch-tested with a large and evolving series of metals. Little has been published regarding many of the metals used in this patch test series. This study encompasses a decade of tests that were performed on patients with a wide variety of skin and mucosal conditions. There was consistency in the application and reading of the patch tests at Mayo Clinic in Rochester. To improve the detection rate of allergic reactions, patch tests were interpreted consistently up to 2 weeks after application of the materials.

We acknowledge that the study has limitations. It is a retrospective study. Only the results of patch testing are included; data describing the outcomes of allergen avoidance were not available from our database. No final diagnosis was documented for more than one-third of our patients.

The interpretation of patch-test reactions remains the biggest obstacle in patch testing with metals in various settings. The only true way to judge relevance would be to determine whether a condition improved after the removal of a suspected allergen; this has been difficult to do, despite some efforts.

Future of the Metal Series

Numerous questions remain unanswered. Which metals should be used for patch testing? Which compounds are more relevant? Which metal preparation (ie, salt and concentration) should be used? How do we differentiate between irritant and allergic patch-test reactions? Are allergies to less-familiar metals more common than previously suspected?

Most important, how do we decide whether an allergic patch-test reaction to a metal is relevant? Is it worthwhile to perform patch testing before the implantation of metal devices in case the patient is allergic to the implanted metal? Should we patch-test a patient who has a loosened metal hip implant and then remove the implant if a positive reaction to its metal is observed? For patients with oral disease, what role does contact allergy to dental crowns, fillings, and bridges have in the development of burning mouth syndrome, oral lichen planus, and stomatitis? Future studies are needed to answer these questions.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: