Neither Skin Tests nor Serum Enzyme-Linked Immunosorbent Assay Tests Provide Specificity for Protamine Allergy

Protamine sulfate administered intravenously (IV) neutralizes the anticoagulant effects of heparin. Despite adverse effects, it remains widely used for this purpose. Untoward responses to protamine vary from urticaria and rash to systemic hypotension, bronchospasm, pulmonary hypertension, cardiovascular collapse, and death [1-3]. Life-threatening reactions to protamine, although extremely rare, entail substantial risk. Allergic reactions to protamine occur sporadically, but arguably with increased incidence in diabetic patients, especially those who receive either neutral protamine Hagedorn insulin or protamine zinc insulin preparations, patients with true fish allergy or a history of previous protamine exposure, and those who have undergone vasectomy [4-8]. Preoperative identification of the protamine allergic patient would allow patient preparation with antihistamines and steroids to reduce the severity of an allergic reaction [9] or selection of an alternative to protamine [10]. Skin tests are the basic tool for testing immunoglobulin class E (IgE)-mediated hypersensitivity [11]. However, they are highly labor intensive and involve exposure to allergen. Enzyme-linked immunosorbent assays (ELISAs) on patient serum can detect minute concentrations of circulating antibodies to an allergen [12]. These tests feature batched analysis and no exposure to antigen [13]. After caring for a series of patients with potential protamine allergy using skin and ELISA testing, we suspected a high false-positive rate of these tests for the outcome variable of hypotension after protamine administration (unpublished observations). Thus, we undertook a limited prospective study of patients undergoing elective cardiac surgery to evaluate the specificity of skin tests and serum ELISA tests. Methods Patients undergoing cardiac surgery prospectively participated with written, informed consent after institutional review board approval. All patients received morphine, 0.1 mg/kg, and scopolamine, 0.006 mg/kg, preoperatively. Upon arrival in a holding area, patients first donated serum, which was frozen at -60 degrees C for subsequent batch analysis via ELISA for antiprotamine IgE, IgE2, and IgG (Protamine Predictor Assay Trademark) by the assay supplier, AlerCHECK, Inc. (Portland, ME). In this process, the test sample is added to protamine-impregnated strips. Any protamine-specific antibodies in serum bind to protamine on the strips. Human anti-Ig, coupled to an enzyme, when added to the preparation, binds to the protamine-specific antibodies. Unbound material washes free of the strip. Addition of a chromogen enzyme substrate then produces a color reaction proportional to the amount of bound enzyme. A color reaction of serum diluted at least 1:10 determined a positive response. The assay featured 100% sensitivity and 94% specificity in a 300-patient cohort (unpublished data, AlerCHECK, Inc.), i.e., a 6% false-positive rate in a 1% true-positive population with no false-negatives. After collection of serum for ELISA, intradermal skin testing on the volar aspect of the forearm occurred prior to the induction of anesthesia. Preservative-free 0.9% saline (S, negative control), d-tubocurarine 100 micro gram/mL (C, positive control), protamine 1 micro gram/mL (P), and protamine 10 micro gram/mL (P10) were injected in random sequence as 0.02 mL via a 27-gauge needle in strict accordance with the technique described by Fisher [14]. Coded syringes permitted double-blind placement and measurement to eliminate operator bias. Measurements of skin induration diameter of 1-mm precision were made 10 min after injection but prior to induction and 30 min after injection soon after induction. In this way, each protamine skin test contained both a positive and a negative within-patient control to determine the threshold induration for a positive test and rule out dermotagraphism [14]. Monitors for surgery included electrocardiogram with multilead ST analysis, pulse oximetry, capnography, airway inflating pressures, and arterial and oximetry pulmonary artery catheters. Sufentanil, 2-3 micro gram/kg, and vecuronium, 0.1 mg/kg, were used for induction of anesthesia. After endotracheal intubation, supplemental sufentanil, up to 8 micro gram/kg, and isoflurane 0.2%-1.5% in oxygen maintained anesthesia. Beef lung heparin, 300-400 U/kg initially, with 5000-U supplements as needed, provided anticoagulation for cardiopulmonary bypass (CPB) as confirmed by duplicate activated coagulation time tests >480 s. Cold sanguinous cardioplegia arrested the heart for surgery. During CPB, anesthesia maintenance included isoflurane 0.2 vol%-1.0 vol% and lorazepam 1-2 mg IV. After termination of CPB, IV protamine sulfate, 1 mg for each 100 U heparin administered infused over 10 min, restored coagulation as guided by return of the activated coagulation time to within 15 s of baseline. A diagnosis of adverse reaction to protamine required, within 20 min of the end of protamine infusion, a >or=to20% decrease in mean arterial blood pressure and one of the following: audible wheezing via esophageal stethoscope, a >or=to10 cm H2 O increase in plateau airway pressure, a >or=to20% increase in mean pulmonary arterial pressure, a >or=to40% decrease in mean pulmonary arterial pressure, or a new skin rash. A rash that was not seen until drape removal would satisfy the last criterion. Fisher's exact test compared skin test and ELISA results, with P < 0.05 considered significant. Standard formulas calculated specificity, and, for the within-patient controls, sensitivity. Based on an expected incidence of clinical adverse response to protamine of <1%, we expected no patient to sustain an adverse response. Thus, a sample size of 32 would permit estimation of test specificity >or=to90%, while a sample size of 64 would permit estimated specificity of >or=to95% [15]. We elected to analyze results after 32 patients and to proceed to 64 patients only if necessary. Results Of 32 participants, aged 21-82 yr, 9 were diabetic, 2 received neutral protamine Hagedorn insulin, and 3 had had vasectomies. Protamine doses ranged from 250 to 500 mg by slow IV infusion. We first used the positive and negative controls (C and S) to define the optimal skin induration threshold at 10 min for skin tests in the patient population studied. The sensitivities and specificities for each millimeter of induration measured at 10 and 30 min, obtained from C and S injections Table 1, indicate an optimal criterion of >or=to8 mm induration at 10 min to denote a positive response. The numerical sum of sensitivity and specificity formed the basis for determining the optimum criterion, since positive and negative predictive values are moot in the absence of true-positives. This selection agrees with respect to both time (10 min) and induration (>or=to8 mm) with Fisher's extensive experience in testing for allergy to anesthetics [14].Table 1: Results of Within-Patient Control Skin Tests(Table 2) displays the frequency of positive skin tests and positive ELISA antibody titers. Four of 32 patients responded positively to intradermal protamine 1 micro gram/mL, whereas 10 of 32 responded to protamine 10 micro gram/mL. Four serum samples could not undergo ELISA testing due to shipment error. All patients displayed IgG at high titers (1:1600 or more dilute). IgE2 and IgE antibodies occurred at 1:10 dilution or greater with surprisingly high overall frequencies of 54% and 46%, respectively.Table 2: Frequency of Positive Results of Skin Tests and Antibodies(Table 3) associates the skin test results with the ELISA test results. The standard (P) and 10-fold concentrated (P10) protamine skin tests underwent separate analyses. The correlations for P and 10P results with IgE and IgE2 ELISA results appear separately. Correlations for IgG are moot, since every patient displayed a high IgG titer. None of the four combinations, P or P10 versus IgE or IgE2, disclosed a significant correlation. No patient developed an adverse response to protamine infusion as defined by the criteria in Methods.Table 3: Frequency of Positive Skin Tests and Positive AntibodiesDiscussion Anaphylactic reactions, although rare, are a major problem because of significant morbidity and mortality. Adverse responses to protamine include anaphylactoid events, such as systemic hypotension with or without pulmonary vasoconstriction [3,16] after too rapid administration [17], as well as true anaphylaxis, i.e., IgG- or IgE-mediated idiosyncratic cardiovascular collapse with or without complement activation unrelated to rate of administration [4,7,9,13,18,19]. Slow administration of protamine can help decrease the incidence of anaphylactoid responses [16,17]. Can one identify patients at risk for an idiosyncratic response to protamine? Previous work focused on demographic factors [6,7,20-22]. Those potentially at risk for anaphylaxis to protamine compose a large population, e.g., patients with prior protamine exposure, those with diabetes mellitus, and those who have undergone vasectomy. Clinicians cannot easily withhold protamine from this large group of patients because of the hemorrhagic consequences. Although substitutes for protamine may become available in the future [10,23,24], these choices may prove too costly to offer all patients at theoretical risk. Likewise, the routine administration of histamine receptor antagonists and corticosteroids to such patients is a wasteful approach and of uncertain protective ability. A test capable of identifying patients truly at risk would permit appropriate allocation of expensive alternatives. With increasing pressure from managed care for "fast track" programs, a rapid, bedside test assumes great importance. Skin testing remains the preferred screening method for individuals at risk for hypersensitivity reactions [11]. It combines low cost with a very low rate of false-negatives, i.e., individuals who test negative rarely react adversely to the allergen. In this study, each patient acted as a positive and negative control to help establish an appropriate positive standard. Induration >or=to8 mm diameter indicated a positive skin test, since the usual 10 mm criterion was insensitive to the standard dilution of curare that should elicit that response [14]. Skin tests also have poor specificity, i.e., many individuals who test positive do not react adversely upon antigen challenge. In this study, 13% of patients had positive skin tests despite having no clinical reaction Table 2, i.e., there was a specificity of 87%. This result corroborates previously reported data [25]. Indeed, comparison of the 87% specificity obtained from intradermal protamine with the 91% specificity obtained from intradermal saline Table 1 reflects the poor specificity of skin testing in general. Would different definitions of positivity improve the results of the current study? Specificity would have improved from 87% to 97% with the more stringent criterion of either 9-mm or 10-mm induration diameter (calculation not shown). However, those choices impose an unknown and unmeasured effect on the test's sensitivity to protamine, since then only 66% and 22% of patients would have tested positive to curare. Similarly, use of the 30-min data would yield specificities of 97%, 100%, and 100% for 8-, 9-, and 10-mm induration diameters, respectively, but with corresponding sensitivities to curare of only 69%, 59%, and 31%. For these reasons, we believe the chosen definitions of positivity should not be altered. Antiprotamine IgE and IgG antibodies have been linked to life-threatening reactions to protamine [7,26]. ELISA testing quantifies titers of circulating antibodies to an allergen. This relatively expensive test requires sophisticated equipment and highly trained personnel. Another sophisticated laboratory test, the radioallergosorbent test, or "RAST," uses radionuclides. Neither test provides rapid results. The false-positive rate of 6% reported by the protamine ELISA test kit manufacturer contrasts starkly with the current results of 46%. Possible explanations for the discrepancy between the manufacturer's data and those reported here include sample degradation during shipment, a change in technique at the manufacturer's laboratory where all samples were analyzed, or peculiar disparities in the patient populations. The 100% incidence of IgG antibody in the current study raises serious doubts about the clinical application of the test. As expected, no patient in this study sustained an adverse reaction to protamine after CPB. While these data can address the specificity, because of the small sample size, they afford no conclusions regarding the sensitivity of tests. Even a cohort twice the current size would have been unlikely to reveal a single patient with an adverse response to protamine. The high incidence of false-positive results using skin tests or ELISA tests illustrates their uselessness to screen patients in "at risk" demographic groups: a positive result is uninterpretable. A negative result may have meaning, but our data cannot address that issue, since this study did not attempt to measure test sensitivity. Until there are more predictive tests, clinicians should treat positive results with skepticism.