"MacroLH": anomalous molecular form that behaves as a complex of luteinizing hormone (LH) and IgG in a patient with unexpectedly high LH values.

Interference in immunoassays from endogenous antibodies directed against peptide hormones is a well-documented phenomenon. It is known to occur for insulin, growth hormone, and especially prolactin, being responsible for most cases of “macroprolactin” (1)(2)(3). However, interference is rare in assays for pituitary glycoprotein hormones, being described only for thyroid-stimulating hormone (4)(5). To date, it has not been described for luteinizing hormone (LH).

We present the case of a female patient, 23 years of age, with a diagnosis of congenital adrenal hyperplasia …

and 172 mg/L) overlapped those detectable in severe bacterial infections.
PCT may be a sensitive marker of invasive bacterial infection (6 ). On the other hand, local bacterial infections, viral infections, and chronic inflammatory processes only slightly increase plasma PCT concentrations (7,8 ). There are few published studies of PCT in common EBV infections. In infectious mononucleosis, the symptoms frequently simulate those of bacterial infection, and CRP does not provide help because CRP values are often increased in this disease, similarly to those found in severe bacterial infections, as confirmed in our study. However, all of our study patients had PCT concentrations within the reference interval for health children (9 ).
In conclusion, in EBV patients under age 5 years, plasma PCT was normal, whereas ESR, WBC count, and CRP concentrations overlapped those found in severe bacterial infections. We acknowledge the need for a study of a different design to assess the utility of a diagnostic test. Further studies of diagnostic accuracy performed prospectively with large samples are required to confirm these findings. To the Editor Interference in immunoassays from endogenous antibodies directed against peptide hormones is a welldocumented phenomenon. It is known to occur for insulin, growth hormone, and especially prolactin, being responsible for most cases of "macroprolactin" (1-3 ). However, interference is rare in assays for pituitary glycoprotein hormones, being described only for thyroid-stimulating hormone (4,5 ). To date, it has not been described for luteinizing hormone (LH).
We present the case of a female patient, 23 years of age, with a diagnosis of congenital adrenal hyperplasia (late-onset 21-hydroxylase defect) and Hashimoto thyroiditis, both conditions under treatment. The laboratory evaluation showed a high LH value (206.0 IU/L) with follicle-stimulating hormone (4.0 IU/L), estradiol (66 ng/L), and prolactin (10 g/L) values within the appropriate reference intervals. Serum human chorionic gonadotropin was undetectable (Ͻ2 IU/L), and the thyroid-stimulating hormone (1.0 mIU/L) and ␣-subunit concentrations (553 ng/L; reference interval, 80 -604 ng/L) were within the appropriate reference intervals. Antibodies against thyroperoxidase were present in high concentrations (1400 kU/L; reference values Ͻ40 kU/L). The patient was not using any LH-stimulating drug, nor had she ever received LH or human chorionic gonadotropin injections. LH was measured by an immunofluorometric assay (in house), and the values were confirmed by an electrochemiluminescent assay (Roche). Serial dilution showed parallelism with the curve obtained with a standard LH preparation. Her serum was subjected to gel-filtration chromatography on a Superdex 200 column (1.5 ϫ 30 cm; Pharmacia) calibrated with the Pharmacia high-molecularweight calibrators, and the elution profile showed that almost all of the LH eluted as a high-molecular-weight form (M r Ͼ250 000; Fig. 1). Recovery after precipitation with polyethylene glycol (6 ) was only 7% compared with a mean recovery of 87.5% (range, 56 -108%) obtained for 23 sera from female patients with LH values Ͼ40 IU/L. Serum application to a protein G-Sepharose column (Pharmacia) showed complete binding of the LH immunoreactivity, which was eluted by lowering the pH to 2.8. Gel filtration in dissociating conditions (0.1 mol/L glycine-HCl, pH 2.8) shifted the LH peak to the expected molecular size.
What called our attention to this case, and prompted additional studies, were the unexpectedly high LH values, with concomitant FSH and estradiol values within the appropriate reference intervals. The etiology and frequency of the phenomenon as well as its relationship with autoimmune diseases remains to be defined. Nevertheless, this new possibility must be considered in the event of a finding of unexpectedly high LH values.

Measurement of Whole-Blood Potassium-Is It Clinically Safe?
To the Editor: Potassium measurement in wholeblood specimens is fast and convenient and is increasingly offered as an adjunct to blood gas analysis. However, the presence of hemolysis in whole-blood samples cannot be determined by such analytical systems at present. There thus is a risk of reporting misleadingly increased whole-blood potassium for unrecognized hemolyzed samples, with potential for subsequent patient misdiagnosis and mistreatment. The real importance of this theoretical concern is unclear. This study describes the prevalence of hemolysis in bloodgas specimens and assesses the subsequent misreporting of whole-blood potassium as a result of hemolysis. It also assesses the importance of low sample volume, age, and gender as risk factors for sample hemolysis.
For 10 days, whole-blood potassium was measured with an AVL Omni 6 blood gas analyzer (Roche Diagnostics) in all arterial blood-gas specimens received by the laboratory. The manufacturer's suggested reference interval for whole-blood potassium is 3.5-4.5 mmol/L. All blood-gas specimens are collected with BD Preset 3-mL draw, dry lithium heparinized syringes (Becton, Dickinson and Company) designed for pH/blood gas and electrolyte determinations. The sample volume and the patient's age and sex were recorded. The specimen was then transferred to a plastic tube and centrifuged at 2000g for 10 min, and the hemolysis (H) index was measured on the supernatant with a 917 clinical chemistry analyzer (Roche Diagnostics). A mean potassium/hemoglobin (K/Hb) ratio of 0.284 mmol/g (with 95% population limits of 0.21 and 0.345 mmol/g) was used to estimate falsely increased potassium measurement attributable to hemolysis (1 ). [Note that the K/Hb values of 28.4 and so forth used in Ref.
(1 ) are incorrect by a factor of 100.] Multiple logistic regression analysis was performed using SPSS, Ver. 11 (SPSS Corporation).
We received 610 samples for blood-gas analysis, of which 60 were Յ0.2 mL in volume and were insufficient for H index measurement. Of the remainder, 18% had a H index Ͼ1 g/L, indicating at least mild hemolysis (14% with a H index of 1-2.5 g/L; 3.6% with a H index of 2.5-5 g/L; 0.4% with a H index Ͼ5 g/L). On the basis of the mean K/Hb ratio (and 95% population limits) from above, the whole-blood potassium measurement error attributable to hemolysis was Ն0.5 mmol/L in 8% (4 -10%) of samples. Table 1 shows the effect of hemolysis on classification of results based on the mean K/Hb ratio. After estimation of the hemolysis effect, 22% (15-24%) of normokalemic samples were downgraded to hypokalemic, and 14% (14 -14%) of hyperkalemic samples were similarly downgraded [8% (8 -