was stopped the patients made a full and uneventful recovery.
The first documented report of haemolytic anaemia, published in the Lancet, came from France. This was a case of immune haemolytic anaemia and acute renal failure in a 50-year-old woman, who was diagnosed in May 1978 (Bournerias and Habibi, 1979). She had had seven episodes of malaise, chills, pain and fever of 2-4 h duration that were accompanied by dark urine and transient jaundice.
During one of the episodes in July 1978, she had had oliguria. At this time, she had a positive Coombs' test and a haemoglobin level of 10 g/dl. Before the episode, she had been treated for an unrelated illness with levomepromazine, diazepam and nomifensine. She made an uneventful recovery on stopping the medication. The serum of the patient demonstrated an antibody that agglutinated red blood cells only in the presence of nomifensine. The authors called for immunologi-cal studies for anti-nomifensine antibodies in patients on long-term treatment.
Another case of acute haemolysis and renal failure (following an overdose of nomifensine) was published the following year (Prescott et al., 1980) (see Table 12.1) and three others from outside the United Kingdom were published in 1981-1982 (Eckstein et al., 1981; Habibi et al., 1981). One of these cases had intravascular haemolysis during treatment with nomifensine (Lyllof et al., 1982).
Although these reports were of concern it was not considered at the time that nomifensine was more liable to cause haemolytic anaemia than other marketed drugs. However, heightened vigilance was recommended and the manufacturer initiated a number of retrospective and prospective immunological studies. These investigations failed to provide support for a cause-and-effect relationship between nomifensine and haemolytic anaemia. Some patients with haemolytic anaemia had a negative Coombs' test, while other patients with a positive Coombs' test did not have haemolysis. Nevertheless, in view of the suspected link between the antidepressant and the blood dyscrasias, haemolytic anaemia was included among the side effects listed in the January 1981 data sheet.
Between 1981 and 1982 there were three more UK cases of haemolytic anaemia reported to the
Department of Health and Social Security (DHSS). They had not been referred to the manufacturer. They occurred among patients who had received a total of 990 000 prescriptions for nomifensine. This suggested an incidence of only about 1 per 150 000 patients, and thus no regulatory action was considered necessary (CSM Update, 1986; Mann, 1988).
These reports did not provide a consistent basis for any general announcement concerning the safety of nomifensine from the company or from a regulatory authority. They placed nomifensine at worst with a group of marketed drugs associated with haemolytic anaemia. This included stibophen, quinidine, paracetamol, penicillin, sulphonamides, tolbutamide, chlorpromazine, tetracycline, cepha-losporins, insulin, rifampicin, hydralazine, streptomycin, triamterene and probenecid for immune haemolytic anaemia, and amongst methyldopa, mefenamic acid, flufenamic acid and levodopa given for autoimmune haemolytic anaemia.
Nevertheless, the company acted on the reports to institute both retrospective and prospective studies in Germany, France, the United Kingdom and Austria, in order to determine potential groups at risk. Between January 1979 and June 1980, 312 patients in these studies who had been treated for more than three months with nomi-fensine were given a Coombs' test, and sera from 220 patients were subjected to intensive immuno-logical investigations. Even with these studies, the results did not prove a causative link with nomifensine. The Coombs' test proved to be inappropriate as a prediction of possible groups at risk amongst nomifensine users. Some patients without haemolysis had a positive Coombs' test and later several patients with haemolytic anaemia were found to have a negative Coombs' test.
In the course of time, supportable evidence for attributing haemolytic anaemia to nomifensine was produced, and in January 1981 this addition to the UK data sheet was agreed: ''Haemolytic anaemia has also been reported in rare cases as has a rise in body temperature". This also appeared in the ABPI Data Sheet Compendium in October 1981.
Concern over the occurrence of haemolytic anaemia and the other serious reactions led to a number of additional immunological investigations, and this work in due course provided further evidence for the immunological basis of the haemo-lytic anaemia reaction (Walti et at., 1983; Miescher, 1985; Salama and Mueller-Eckhardt, 1985).
Salama et at. (1984) demonstrated a nomifen-sine-dependent antibody that reacted exclusively to its ex vivo antigen (fresh serum of a volunteer who had taken a therapeutic dose of the drug), but not to nomifensine itself. The investigators later showed an "extraordinary heterogeneity" of antibody response following the ingestion of the antidepressant. Of 19 samples only five were primarily reactive to nomifensine. The majority reacted in the presence of one or more metabolite and ex vivo antigens, indicating specificity for an unidentified early or late metabolite.
All samples belonged to the IgG or IgM class or both and were capable of activating complement. At least one sample had two nomifensine-depen-dent red blood cell antibodies, while one had platelet antibodies. The latter explained the occurrence of purpura alongside the haemolysis. It is of interest also that seven of the 19 patients had signs of transient renal insufficiency, while six had increased levels of serum transaminase (type not specified; Salama and Mueller-Eckhardt, 1985).
Previously (in September 1978, published April 1979) the data sheet had been amended to draw attention to the association of nomifensine with fever. There had been several reports of this in Germany and five reports were submitted to the UK manufacturer in 1977. The data sheet stated that there had been ''rare cases of rise in body temperature which returned to normal when the drug was withdrawn".
The data sheet of 1981 also drew attention to the association of nomifensine with changes in liver enzymes by stating that:
In rare cases, increases in liver enzymes (serum transaminases and alkaline phosphatase) have been observed.
As a result of receiving four reports of haemolytic anaemia in 1978-1979, the manufacturer undertook the following actions:
• Full investigation of each case report. The normal company operating procedure involved acquiring full information on each case from the prescribing doctor, if necessary visiting the doctor to discuss the case and being accompanied on such visits by medical personnel from the central drug safety department of the company headquarters.
• All cases to be reported to the parent company and the UK DHSS.
• Re-appraisal of all preclinical work and clinical trials to see whether there was any evidence of blood dyscrasias. None was found.
• Retrospective and prospective immunological studies. These produced no consistent results related to the clinical use of the drug.
• Sales representatives to be informed of publications and investigative activities in order to respond appropriately to enquiries.
• Data sheet changes with international agreement relating to fever, haemolytic anaemia and the liver
Was this article helpful?