Does a negative direct antiglobulin test exclude warm autoimmune haemolytic anaemia? A prospective study of 504 cases

Serological evidence of warm-type autoimmune haemolytic anaemia (AIHA) is generally obtained by a direct antiglobulin test (DAT). With few exceptions, a positive DAT is indicative for the presence of autoantibodies (Gehrs & Friedberg, 2002). However, DAT-negative AIHA is suspected to constitute up to 5% of all warm-type AIHAs. It can occur by several possible mechanisms, including sensitisation by a small amount of IgG that falls below the detection threshold and both IgA and IgM antibodies that remain undetected by polyspecific antiglobulin reagents (Garratty, 1993). A highly sensitive gel technique has been introduced to overcome these problems (Nathalang et al, 1997). Accordingly, a negative DAT is now often interpreted as absence of autoantibodies and thus, as more or less preclusive for the diagnosis of AIHA.

We conducted a prospective study enrolling in- and out-patients in whom immune-mediated red blood cell (RBC) destruction was to be excluded (n = 1388). An indirect antiglobulin test (IAT) and DAT including anti-IgG, -IgM, -IgA, -C3d and -C3c specificities, respectively, were performed using the gel technique. Acid elutions were subsequently performed on RBCs from all patients who were non-reactive in IAT and DAT (n = 504).

Out of the 504 eluates performed, 481 were non-reactive, but 23 (4·6%) contained RBC-specific antibodies. Patients’ details are given in Table I. Two groups of patients were identified: patients with previously undiagnosed AIHA (15/23) and patients with allospecific antibodies (5/23). All AIHA patients were found to be anaemic, with one exception (Patient 12) in whom a haemoglobin concentration at the lower end of the normal range was observed. A secondary cause of AIHA was determined in six patients (Patients 1–6). One year after the initial diagnosis, five patients had responded to steroid treatment and seven displayed at least one positive DAT upon subsequent examinations (data on treatment not available).

Alloimmunisation because of recent RBC transfusions was identified as a further cause for the serological constellation of negative DAT and positive RBC eluate. Two patients (Patients 19 and 20) displayed a delayed serological transfusion reaction (DSTR), two others (Patients 21 and 22) were diagnosed with a delayed haemolytic transfusion reaction (DHTR). A free alloantibody became detectable over the course of time in all four patients. The remaining four patients did not have any evidence of haemolysis.

This, to our knowledge, is the first prospective study demonstrating that even with an improved, monospecific DAT, employing the gel technique, a small, but not negligible number of AIHA patients will remain undetected. The combination of a positive eluate from DAT-negative RBCs was described over 30 years ago (Gilliland et al, 1971). It has been attributed to the fact that, upon elution from several millilitres of RBCs, the concentration of the autoantibodies reaches the critical amount necessary to induce RBC agglutination. This is not unlikely, as Mollison and Hughes-Jones (1967) have demonstrated that the lowest concentration of antibodies capable of bringing about RBC destruction was 1/50th of the lowest concentration detectable in IAT (although, at that time, it was performed as a tube test). In addition to the amount of antibodies, other factors, e.g. the orientation of immunoglobulins bound to the RBC surface, have also been discussed (Garratty, 1989).

Two major conclusions can be drawn from our study. The DAT must be interpreted in conjunction with clinical and other laboratory data to avoid erroneous conclusions; it can be expected to be false-negative in up to 3% of AIHA patients. Elution of RBC antibodies is a valid additional procedure to clarify whether autoantibodies are present in DAT-negative patients.

Besides the detection of autoantibodies, alloantibodies because of RBC transfusions may be detectable in the early stages of immunisation by RBC elution. Accordingly, serological evaluation of suspected haemolytic transfusion reactions should involve RBC eluates.