sign in sign up
<<
Home , African iron overload, Iron overload

Letters 937

Hepatocellular carcinoma and African We entirely support the authors in their (HCC) and the tox- overload urge that prospective studies should be icity of in . undertaken to examine the possible role of High quality cancer epidemiological data EDITOR,-Gangaidzo and Gordeuk (Gut African overload in the pathogenesis of HCC. are difficult to come by in Africa. With regard 1995; 37: 727-30) postulate that iron over- However, at the village level, in the requisite to the estimate ofDrs Walker and Segal offive load may be a risk factor for hepatocellular studies, major difficulties will be encountered admissions for HCC per 100 000 population Gut: first published as 10.1136/gut.38.6.937-a on 1 June 1996. Downloaded from carcinoma (HCC) - a subject of obvious regarding participation, especially regarding annually in rural South Africa, we note that interest to workers in Africa. It is especially so taking and the fear of AIDS. A compli- hospital based studies may be unreliable to one of us (AW) who, originally, in 1953, cating factor for HCC, as Kew and associates9 estimates of countrywide incidence rates.' propounded the iron overload hypothesis. have shown, is that the infection rate of HBV, HCC is a common and important condition This sought to explain the abnormal iron an important aetiological factor, varies not in Africa. At least one authority has observed, deposition () present in many only from region to region, but from village to with regard to HCC, that 'some of the docu- Africans, caused by a high adventitious intake village. Additionally, there will be the usual mented incidences are almost certainly of the element arising from food preparation difficulties in seeking to resolve whether iron underestimates, perhaps by as much as 50 in iron utensils.' overload, if implicated, is a causative or an percent in some countries'.2 A contemporary In comment, firstly, the above authors state associated factor. Not least of perplexities textbook of in the tropics that HCC is probably the commonest malig- is the situation whereby a given parameter, offers an adjusted rate for HCC of 113 per nancy occurring in men worldwide. Actually for example, serum concentration, 100 000 per annum in Shangaan men inhab- it ranks eighth in numerical importance on a can have a differing significance according to iting the border areas of Mozambique, worldwide basis, accounting for 5-3% of new the context. As a recent example of this Zimbabwe, and South Africa.2 In Harare, cancers in men and 2-6% in women.2 phenomenon, at Belfast and Toulouse, there Zimbabwe, HCC is the most common malig- Next, it is insufficiently appreciated that its is the same average serum cholesterol value nancy in men in the 1990s (Harare Cancer incidence in men in Africa is highly variable, but a fourfold difference in mortality from Registry, 1995). In the internal medicine ranging from 47-9 per 100 000 world popula- coronary disease.'0 wards of one of us (ITG) at Harare Central tion, in Mali,3 to 7-5 in Uganda,4 and 6.4 in Finally, should iron overload be meaning- Hospital, HCC was the most common cancer South Africa.5 In agreement with the latter, fully incriminated, the authors say 'it is diagnosis in men in 1993-1994 and recently, in three rural widely separated eminently preventable through changing the accounted for 3-2% of admissions. Among hospitals, responsible for the needs of about practices of preparation and consumption of those of us who regularly provide medical 400 000 Africans, we found an average offive traditional beer, and it is treatable by phlebot- care to rural Africans (constituting more than admissions of men for HCC per 100 000 omy therapy to remove iron from the body'. 80% of the Zimbabwean population), there is annually. Interestingly, the latter rates are In our view, both ofthe suggested procedures, general agreement that HCC represents a much the same as those of Afro-Americans, in the rural context, are non-starters. major health problem. We have seen 'few 50-6.6 per 100 000.3 The citing of 100 cases To reiterate, while we are unconvinced of more depressing tasks than caring for a per 100 000 in Mozambique is inapposite as the clinical importance of iron overload, the patient with this particular malignancy, no current study ofthe painstaking type made issue must be resolved - that is, whether it is and there is no satisfactory treatment'.3 in Uganda,4 has been pursued in or is not of public health significance within The question of whether iron overload in Mozambique - a country at war for 20 years. the context of impoverished Africa. Africa is 'noxious', seems to recapitulate the We know of no present day excessively high A R P WALKER discussion of 30 years ago about whether iron rates for HCC in African countries south of Human Biochemistry Research Unit, overload is damaging in people of European Department of Tropical Diseases, origin with hereditary haemochromatosis. As the Equator. School ofPathology ofthe of iron we learn more about states of excess iron, it is Turning now to the noxiousness University ofthe Witwatersrand, and the http://gut.bmj.com/ overload, in the original hypothesis,' also South African Institutefor Medical Research, our opinion that the weight ofevidence points on 296 Johannesburg, to the conclusion that iron overload of what- in a subsequent local study based South Africa postmortem examinations at Baragwanath ever cause (HLA linked hemochromatosis, Hospital, Soweto, Johannesburg,6 the patho- ISIDOR SEGAL transfusions, ineffective erythropoiesis, or Gastroenterology Unit, dietary iron in Africa) is potentially toxic, genicity of iron overload was doubted, for no Baragwanath Hospital high constant correlation was found between the and University ofthe Witwatersrand, and that medical practitioners are obligated to degrees of and and the Johannesburg, work to prevent and treat iron overload in amount ofiron pigment in the . However, South Africa their patients.4 In a series of careful patholog- ical studies 35 years ago, Professor T H on September 28, 2021 by guest. Protected copyright. later, at the same hospital, as recently detailed 1 Walker ARP, Arvidsson UB. Iron 'overload' in by Lynch,7 associations were reported of the South African Bantu. Trans R Soc Trop Bothwell and colleagues at the University siderosis with , , oeso- Med Hyg 1953; 47: 536-48. of the Witwatersrand provided convincing 2 Tomatis L. Cancer: causes, occurrence and control. evidence that dietary iron overload directly phageal cancer, and . The find- International Agency for Research on Cancer, ings in these studies, however, made over 20 IARC Scientific Publications No 100. Lyon: causes cirrhosis.5-7 Interestingly, in 1960 years ago, have not been currently confirmed. World Health Organization, 1990: 59. Bothwell reported that in Africans the hepatic More to the point, over the decades, no 3 Parkin BM, Muir CS, Whelan SL, et al. Cancer iron threshold above which portal fibrosis and incidence infive continents. Vol 6. Lyon: IARC, are to be is a concen- attempt has been made at the village level to 1992: 930. cirrhosis likely present learn whether iron overload contributes 4 Wabinga HR, Parkin DM, Wabwire-Mangen F, tration of 360 ,umol/g dry weight5 (normal is and that Mugerwa JW. Cancer in Kampala, Uganda, in less than 30 ,umol/g dry weight). Twenty six significantly to morbidity mortality; in the era of is, is it disadvantageous to 1 per 100, or 1 per 1989-91: changes in incidence years later Dr L Powell's group in Australia AIDS. Int_J Cancer 1993; 54: 26-36. for 1000? 5 Sitas F, Pacella R. Cancer regist"y ofSouth Africa, reported an almost identical threshold In an investigation made in 1985 by 1989. Johannesburg: South African Institute hepatic damage among white subjects with Gordeuk et a18 in Zimbabwe, it was estimated for Medical Research, 1994: 33. hereditary hemochromatosis.8 Support for the 6 Higginson J, Gerritsen Th, Walker ARP. that there were about 80 000 cases of severe Siderosis in the Bantu of . association between iron overload. In the same year, it was stated Am JPathology 1953; XXIX: 779-815. and cirrhosis is provided by a study con- that in that country there were 14 587 cases of 7 Lynch SR. Iron overload: prevalence and impact ducted in the late 1980s at a mission hospital on health. Nutr Rev 1995; 53: 255-60. in Swaziland: nine of 29 consecutive adults malaria, 2956 of tuberculosis, and 144 of 8 Gordeuk VR, Boyd RD, Brittenham GM. cholera. Surely, were iron overload as patho- Dietary iron overload persists in rural sub- undergoing diagnostic liver had gnomonic as is conjectured, there should be a Saharan Africa. Lancet 1986; i: 1310-3. hepatic iron concentrations over 360 ,ug/g plethora of evidence incriminating it. This 9 Kew MC. Chronic B virus infection dry weight and seven of these subjects (78%) and hepatocellular carcinoma in Africa. SAfrJ fibrosis or cirrhosis on does not seem to be the case. Sci 1992; 88: 524-8. had either portal Gangaidzo and Gordeuk regret that in 10 Evans AE, Ruidavets J-B, McCrum EE, et al. histological examination.9 Africa, only about half of the cases of HCC Autres pays, autres coeurs? Dietary patterns, Two recent studies that we have conducted risk factors and ischaemic heart disease in examine the noxiousness of dietary iron over- are explicable on the basis of HBV infection. Belfast and Toulouse. Q Jf Med 1995; 88: It must be recognised that apart from the 469-77. load in data sets that span the historical spec- cause and effect of classic deficiency diseases, trum of this disease. Firstly, we analysed data the above proportion or less is common from the original study of iron overload in with multifactorial diseases - dental caries, Africans, conducted by Dr A S Strachan coronary heart disease, and certain cancers. Reply between 1925 and 1928 at Johannesburg In brief, the balance of knowledge and igno- General Hospital, to determine if this form of rance regarding the causation of HCC is the EDITOR,-We thank Drs Walker and Segal for iron loading may be associated with deaths usual. their perspectives on the prevalence of from HCC or tuberculosis.'0 In the original 938 Letters study, necropsies were performed on 604 9 Friedman BM, Baynes RD, Bothwell TH, corticosteroid treatment and it is shown3 4 that Gordeuk VR, MacFarlane BJ, Lamparelli RD, adult blacks from southern Africa and one of et al. Dietary iron overload in southern African these tests, performed during or immediately five grades of hepatic and splenic iron was blacks. SAfrMed J' 1990; 78: 301-5. after corticosteroids have poor ability in dis- assigned based on the Prussian blue reaction 10 Gordeuk VR, MacPhail AP, McLaren CE, criminating the clinical outcome. In our obtained when a piece of tissue was dipped Deichsel G, Bothwell TH. Associations of opinion the French investigators were unable

iron overload in Africa with hepatocellular Gut: first published as 10.1136/gut.38.6.937-a on 1 June 1996. Downloaded from in a mixture of potassium ferrocyanide carcinoma and tuberculosis: Strachan's 1929 to show a prognostic ability of sedimentation and hydrochloric acid.1 1 Using logistic regres- thesis revisited. Blood 1996; 87: 3470-6. rate and albumin because at least 64 patients sion models, we found that iron overload 11 Strachan AS. Haemosiderosis and haemochroma- among 167 included were treated with tosis in African natives with a comment on the was strongly associated with the findings of aetiology of haemochromatosis. [MD thesis]. corticosteroids at the entry of the study cirrhosis (p<00001), HCC (p=0.0002), Glasgow: University of Glasgow, 1929. (placebo group of the trial about usefulness of and tuberculosis (p<0 0001).10 Secondly, prednisolone treatment in patients in clinical we reviewed all 320 diagnostic remission after prednisolone treatment for an specimens processed at the University of acute attack). Zimbabwe from 1992 to 1994. HCC was Analysis ofbiological variables in We agree, however, with Sahmoud's con- present in 19% of the evaluable specimens, Crohn's disease clusions about the necessity of identifying cirrhosis in 21%, and high grades of iron in patients with different risks of relapse in qui- 19%. We found significant associations EDITOR,-We wish to comment on the paper escent Crohn's disease. In our study, at the between the presence ofiron overload and the by Sahmoud et al (Gut 1995; 37: 811-8) end of 18 months of follow up, the predicted histological diagnoses of cirrhosis and HCC. where the authors suggested the following proportion of patients with relapse was 13% We do not completely share the pessimism features: age, duration of disease, interval and 31% in the subgroup with normal labora- of Drs Walker and Segal with regard to chang- since previous relapse, and colonic involve- tory tests and favourable and unfavourable ing the method of preparation of traditional ment as powerful prognostic factors to predict clinic characteristics respectively, and 87% in beer and to instituting a phlebotomy pro- relapse in quiescent Crohn's disease. We also patients with abnormal tests and unfavour- gramme. The iron drums that are now used to followed up for 18 months 107 patients with able characteristics. Thus risk ofrelapse varies prepare traditional beer replaced clay cooking Crohn's disease in clinical remission1 and we substantially among different subgroups of pots around the turn of the century, but these used laboratory tests enhanced by clinical patients with clinical remission; prediction of clay utensils are still used in most rural com- characteristics for predicting relapse. this risk is helpful not only for a correct design munities to prepare food and other forms of Interestingly, our results about clinical char- of clinical trials, but also, in clinical practice, beverages. It seems feasible to us to encourage acteristics were, for some aspects, similar to for timing subsequent clinical visits and for the use ofthese clay pots for the preparation of those obtained by the French group. A Cox selecting groups of patients where preventive traditional beer in place ofthe newer and more regression analysis indicated interval since pre- treatment could be justified. convenient, but probably more dangerous, vious relapse, previous surgery, location (ileum C BRIGNOLA C BELLOLI iron drums. For the past two years our research versus non-ileum), and occurrence of postsur- G DE SIMONE team has conducted a rural based study of gical clinical recurrence as predictors of M CAMPIERI dietary iron overload. We have been struck by relapse; however an intercorrelation was found P GIONCHETTI the level that we these characteristics and A BELLUZZI overwhelming of cooperation among particularly Istituto di Clinica Medica e Gastroenterologia, have been able to obtain through close and between previous surgery and location (but Policlinico S, regular contact with the rural communities. simultaneous Cox regression indicated that Orsola, V, Despite the need for venesection and a high surgery would be more predictive than loca- Massarenti 9, 40138 Bologna, awareness of the problem of HIV, it has been tion). Thus two groups of risk of relapse Italy unusual for subjects to refuse to take part. according to clinical characteristics were In summary, we believe the available data defined: (a) favourable group consisted of 1 Brignola C, Iannone P, Belloli C, De Simone G, http://gut.bmj.com/ in Bassein I, Gionchetti P, et al. Prediction of point to both HCC and dietary iron overload patients clinical remission after last relapse relapse in patients with Crohn's disease in as major heath problems in rural Africa. for at least 24 months or patients with less than remission: a simplified index using laboratory There is a strong body of evidence to suggest 24 months but who had undergone surgery tests, enhanced by clinical characteristics. an association between the two conditions. without subsequent post surgical clinical European J7ournal of Gastroenterology and 1994; 6: 955-61. Major initiatives are needed to combat these recurrence; (b) unfavourable group consisted 2 Brignola C, Campieri M, Bazzocchi G, Farruggia diseases beginning in the communities where ofpatients in clinical remission for less than 24 P, Tragnone A, Lanfranchi GA. A laboratory the people live. months not included in the favourable group. index for predicting relapse in asymptomatic patients with Crohn's disease. Gastroenterology on September 28, 2021 by guest. Protected copyright. INNOCENT TICHAONA GANGAIDZO In the French study surgery was not identi- 1986; 91: 1490-4. Department ofMedicine and the fied as a prognostic factor; this could be 3 Landi B, N'Guyen Anh T, Cortot A, Soule JC, Clinical Epidemiological Unit, explained because in that study quiescent dis- Rene E, Gendre JP, et al. Endoscopic monitor- University ofZimbabwe School ofMedicine, ease induced by surgery seems to be excluded ing of Crohn's Disease treatment: a prospec- PO Box A178, tive, randomized clinical trial. Gastroenterology Avondale, and ileal location (and for this reason also 1992; 102: 1647-53. Harare, previous resection) is present in a small num- 4 Brignola C, De Simone G, Belloli C, Iannone P, Zimbabwe ber of patients in comparison with colonic Belluzzi A, Gionchetti P, et al. Steroid treat- VICTOR ROLAND GORDEUK location. In our experience duration of illness ment in active Crohn's disease: a comparison Department ofMedicine, between two regimens of different duration. George Washington University Medical Center, was not a useful predictor of relapse; we did Aliment Pharmacol Ther 1994; 8: 465-8. Washington DC, not study age of patients. USA In the study by Sahmoud et al biological indicators such as albumin and erythrocyte Reply 1 Sackett DL, Hayes RB. Bias in analytic research. sedimentation rate did not result in predicting Jf Chron Dis 1979; 32: 51-63. 2 Kew MC. Hepatocellular carcinoma: epidemiol- the course of the disease; in contrast, in our EDrTOR,-Thank you for referring the interest- ogy. In: Watters D, Kiire CF, eds. experience, laboratory tests seem to be ing comments ofBrignola et al to us conceming Gastroenterology in the tropics and subtropics. extremely useful for this aim. In a previous our article. The Italian group has considerable London: Macmillan, 1995: 184-6. 3 Cook GC. Hepatocellular carcinoma: one of study2 a prognostic index based on laboratory experience in investigating the value ofbiologi- the world's most common malignancies. tests was proposed. In our last study' the cal parameters in Crohn's disease and has pub- Q J Med 1985; 57: 705-8. ability of this index to predict relapse was lished several research results in that field. One 4 Gordeuk VR, McLaren G, Samowitz W. validated and a simplified application was of their earliest suggested a prog- Etiologies, consequences and treatment of publications' iron overload. Crit Rev Lab Clin Med 1994; 31: proposed; patients with at least one of the nostic index for predicting relapse in quiescent 89-133. laboratory tests changed (a, acid glycoprotein Crohn's disease patients using the baseline 5 Bothwell TH, Bradlow BA. Siderosis in the >130 mg/dl, or a2 globulin >9 0 gm/l or blood values of al glycoprotein, a2 globulin, Bantu. A combined histopathological and chemical study. Arch Pathol 1960; 70: 279-92. erytrocyte sedimentation rate >40 mm/I h, or and the erythrocyte sedimentation rate. This 6 Bothwell TH, Isaacson C. Siderosis in the all three) presented within 18 months a relapse was based on data from 41 patients who had Banns. A comparison of incidence in males rate of 75% whereas in patients with normal been in remission for at least six months. The and females. BMJt 1962; i: 522-4. tests it was 1 In our study nine among 18 same study suggested that the blood values of 7 Isaacson C, Seftel HG, Keeley KJ, Bothwell TH. 3%/. Siderosis in the Bantu: the relationship false-negative (normal laboratory tests with C-reactive protein and al antitrypsin were of between iron overload and cirrhosis. J Lab subsequent relapse) were represented by borderline significance and the predictive value Clin Med 1962; 58: 845-53. patients with previous remission <3 months of some other biological variables were much 8 Bassett ML, Halliday JW, Powell LW. Identification of homozygous hemochromato- who had recently stopped corticosteroid treat- less important, namely the haemogiobin con- sis subject by measurement of hepatic iron ment. Usually, however, values of laboratory centration, the white blood cell count, serum index. HepatologAy 1986; 6: 24-9. tests are rapidly normalised by a period of iron, albumin, and Sy globulin blood values.