Limitations of the paired t-test for evaluation of method comparison data.

patients were referred to an oncologist or gynecologic oncologist with a suspected diagnosis of trophoblast disease or postgestational choriocarcinoma. Four of the six cases received multiple courses of methotrexate chemotherapy, and one of the six received in addition EMACO chemotherapy, with no major quantitative reduction in false-positive hCG results. After chemotherapy, two patients underwent a hysterectomy and one other patient underwent an oophorectomy, all without a major reduction in measured hCG concentrations. One patient developed type 1 diabetes as a complication of the chemotherapy and became comatose. All therapies came to a halt with the finding by the hCG Reference Service that the persistent hCG results, the sole basis for treatment, were in fact false-positive or “phantom” hCG. We have now heard that in two of our earliest phantom hCG cases (tested in Spring 1998), the false-positive hCG results eventually, after 10–14 months, subsided. The cases came from all parts of the United States (one from the West Coast, three from the Midwest, and two from the Northeast). These cases of phantom hCG found their way through word of mouth to the hCG Reference Service, a new, unadvertised facility, in a 9-month period. We wonder how many other similar cases may exist. Four of the six false-positive cases had been detected and followed with the Abbott Diagnostics AxSym hCGb test (Table 1), and one of six had been followed with the sister assay, the Abbott Diagnostics IMx hCGb test, which uses the same antibody/chemical set. Thus, five of the six cases were detected with this one type of assay. We do not know if this type of assay, among the .40 quantitative hCG test sold in the US (4 ), is particularly prone to false-positive results. Laboratory directors and managers need to be aware of this potential problem, especially if they are performing the AxSym or IMx hCGb type test. They need to be available to help physicians quickly exclude or identify phantom hCG, which can be done by simply running quantitative urine hCG tests. In phantom hCG cases, no hCG immunoreactivity may be detected (,5 IU/L) in urine samples. Other simple methods to exclude phantom hCG are to test serum samples with competitive hCGb RIAs (which do not detect phantom hCG), or to demonstrate nonlinearity in dilutions of the serum samples in an hCG immunoassay (phantom hCG may give grossly nonlinear results). Alternatively, help can be sought from the hCG Reference Service.