In the news

Both the US National Institutes of Health and a Health Technology Assessment review in the UK have recommended the introduction of population antenatal screening of prospective parents for cystic fibrosis carrier status (see Cystic fibrosis summary for more information). These recommendations have not met with universal approval. Among the unanswered questions is the best screening method: couple screening, in which both parents supply a sample at the same time, and two-step screening, in which only the mother supplies a sample initially and her partner supplies a sample for testing if she proves positive. One of the criteria for a successful screening programme is accurate recall of risk status by the people who are tested. Marteau et al have surveyed couples participating in trial screening programmes to see whether those who tested negative accurately recalled their residual risk three years later [Marteau, T.M. et al. (1999) Am J Obstet Gynecol 181, 165-169 (Abstract)]. They found that those who underwent couple testing were 4.5 times more likely to recall accurately that they were unlikely (rather than definitely, likely, definitely not, or don't know) to be CF carriers. The authors suggest that the best approach from the psychosocial point of view is complete couple screening and with full disclosure of results, that is, samples are taken from both parents at the same time, both are tested, and both parents are told their result. This differs from the usual method of couple screening, in which the father's sample is taken at the same time as the mother's but only tested if the mother's is positive. Couple screening avoids the anxiety that results from the couple having to wait for the father's test result if the mother has been told that her test is positive.

Comment: The approach to screening is an important parameter because it affects both the financial and the human costs. The approach suggested by Marteau et al would substantially increase the financial cost of the screening programme, as the single most important factor in determining cost is the cost of the DNA test itself, and such tests would have to be performed on both partners of every couple who accepted the offer of antenatal testing, rather than on all mothers and only approximately 5% of fathers (that is, the partners of the 5% of women who tested positive). These costs would probably prove prohibitive. Marteau et al's approach would, however, seem to involve the lowest human costs, both in terms of anxiety, and accuracy of recall of the meaning of the test result. It would also mean that if either parent were involved in a subsequent pregnancy, he or she would not need to be re-tested.

Population screening for prostate cancer, using prostate specific antigen (PSA) and digital rectal examination, is controversial. In the UK, it has not been recommended because of the lack of good preventive options for those who test positive, and the fact that prostate cancer has a late age of onset and often progresses slowly. The balance could perhaps shift in favour of screening for a subset of men who, because of a family history of the disease, were at significantly higher risk. Gronberg et al have used records from the Swedish Cancer Register to estimate the cumulative age-specific risks of prostate cancer in the sons of 5402 fathers diagnosed with the disease [Gronberg, H. et al. (1999) Cancer 86, 477-483 (Abstract)]. The age at which cancer was diagnosed in the father strongly affected the risk for his sons: the relative risks were 2.68, 1.93 and 1.56 for sons of fathers diagnosed under 70 years, 70-79 years and over 80 years, respectively. In families with at least two cases of prostate cancer (that is, the father and at least one son), the cumulative risk of cancer was approximately 5%, 15% and 30% by the age of 60, 70 and 80 respectively, compared with 0.45%, 3% and 10% in the general population.

Comment: On the basis of their data, the authors recommend surveillance for prostate cancer (by PSA and digital rectal examination), beginning at age 50, in men who have at least two close relatives with the disease. However, although these men are certainly at higher than the population risk of the disease, the factors that militate against regular surveillance still apply. More information is needed about mortality from prostate cancer in men with a family history of the disease, and whether regular surveillance could reduce mortality.

Using information on 550,000 primary cancers in the Swedish Family Cancer Database (an essentially complete record for the whole country), Hemminki and Vaittenen have calculated the familial risks for all the common cancers by comparing the age-adjusted incidence rate of cancer in the offspring of probands with the incidence rate for the same cancer in the whole population [Hemminki, K. and Vaittinen, P. (1999) Eur J Cancer 35, 1109-1117](see Cancer summary for further information on the genetics of cancer susceptibility). The familial relative risks (FRR) calculated in this way were mostly around 1.5–3, in general agreement with previous studies on smaller populations. Exceptions included cancer of the testis (FRR 7.4) and thyroid (FRR 8.3). The proportions of familial cases were also calculated, by dividing the numbers of parent-offspring pairs sharing cancer at a particular site by the total number of offspring with that cancer. For the offspring of male probands, the highest proportions of familial cases were found for prostate (11.5%), lung (4.9%) kidney (2.5%) and colon cancer (2.5%). For the offspring of female probands, the highest proportions were for breast (8.7%), uterine (5.0%) ovary (2.7%) and colon cancer (2.7%). The authors of the study and an accompanying Editorial [Easton, D. (1999) Eur J Cancer 35, 1043-1045] comment that the relatively young age range of the offspring (all born since 1941) will affect the results, for example if the age profile for the incidence of familial cancer is different from that of sporadic cancer at the same site. So for breast cancer, for example, the FRR of 1.8 might be expected to fall as the population ages. In contrast, the FRR for ovarian cancer, 2.8, appears to be fairly constant with age.

Comment: The overall conclusion from the study is that, although almost all the common cancers appear to have an inherited component, it is generally quite small. The challenges of identifying the genes responsible are likely to be correspondingly large. Even if/when these genes are found, it is possible that genetic testing for cancer risk will not yield useful information except for a small fraction of families who have a very strong family history of a particular malignancy.  

De Stefano et al report a retrospective study to compare the risk of recurrent spontaneous deep vein thrombosis (DTE) in people with different genotypes for factor V and prothrombin, two factors in the coagulation cascade [De Stefano, V. et al (1999) N Engl J Med 341, 801-806 (Abstract)] (see Hereditary thrombophilia page for references to background information). Patients heterozygous for the factor V Leiden mutation and the prothrombin G20210A mutation had a relative risk of 2.6 for recurrent DTE and 3.7 for spontaneous recurrent DTE compared with either people with neither mutation or people with only the factor V Leiden mutation (spontaneous recurrence means recurrence in the absence of known risk factors such as oral contraceptive use or prolonged immobilisation). De Stefano and colleagues conclude that double heterozygotes are "candidates for lifelong anticoagulation" after a first episode of DTE.

Comment: Public health considerations do not at present support the authors' recommendations. Anticoagulant treatment itself carries significant risks, which are likely to be higher than the absolute risk of mortality from recurrent DTE in double heterozygotes. Genetic testing for all patients suffering from a first episode of spontaneous DTE would be extremely expensive and the information would be unlikely to aid in their management.

In an Editorial in the BMJ, Haddow and Bradley argue that several questions remain to be answered before population screening for hereditary haemochromatosis could be justifiably introduced [Haddow, J.E. and Bradley, L.A. (1999) BMJ 319, 531-532]. Although the disease is quite common (around 1 in 250 people of white European origin are homozygous for C282Y, the major disease-causing mutation in the HFE gene) and a safe and effective treatment (regular phlebotomy) is available, there is insufficient information about the relationship between mutation status and the severity of disease. A significant percentage of people who are homozygous for C282Y have few or no symptoms and may live on into old age without suffering major ill health. There is a risk that such people could be labelled as "ill", and subjected to medical surveillance and treatment that would be not only costly and unnecessary but even have harmful social or psychological consequences. Even assuming there were sufficient evidence to justify screening, there is insufficient information to decide what population group(s) should be screened, when, and by what method. Studies are continuing in attempts to answer some of these questions. In a workplace-based study in the US, McDonnell et al found that, of 6000 employees (80% of whom were under the age of 50), 1653 accepted an offer of testing for haemochromatosis by a transferrin saturation test, and 1450 of these people also agreed to genetic testing [McDonnell, S.M. et al. (1999) Am J Med 107, 30-37 (Abstract)]. 13 people were found to have haemochromatosis on the basis of transferrin saturation (a frequency of 1 in 127) but only 5 had symptoms of iron overload. 8 of those with haemochromatosis were homozygotes or compound heterozygotes for known haemochromatosis mutations, but 4 were genotypically normal. Moreover, 82 people with putative haemochromatosis genotypes (including 2 homozygotes for the major disease-associated mutation) had no symptoms of disease, though the relatively young age of the population and the predominance of women (83%) may affect these results. In a study in Australia, Olynyk et al. [(1999) N Engl J Med 341, 718-724] followed a population of 3011 unrelated white adults over a period of four years, determining their genotype for the HFE gene and measuring serum transferrin saturation and ferritin levels. 16 people (0.5%) were found to be homozygous for the C282Y mutation but although all showed at least some evidence of iron excess in their serum (transferrin saturation or ferritin), about half of them showed no clinical features of haemochromatosis and 25% had normal levels of serum ferritin over the whole four-year period. In addition, 11 people with elevated serum transferrin and ferritin levels were not C282Y homozygotes.

Comment: Haddow and Bradley point out that decisions on screening programmes must be based on the balance between doing good and doing harm. It appears that there is still insufficient evidence to decide where that balance lies in screening for haemochromatosis. However, Tavill, in an Editorial in the New England Journal of Medicine, commenting on the paper by Olynyk et al, takes a more positive view, stating that "the message is clear: most persons who are homozygous for the C282Y mutation have iron overload, and this mutation can be detected reliably by measuring transferrin saturation".

There has been considerable debate about the potential role of general practitioners in genetic screening programmes. Doubts have been raised, for example, about their degree of interest and about their level of genetic knowledge. Leggatt et al have carried out a pilot study in which a postal questionnaire was used to identify people in one general practice who were at a significantly higher risk of breast or colorectal cancer on the basis of their family history [Leggatt, V. et al. (1999) BMJ 319, 757-758 (18 Sept issue general practice section)]. Of 2265 people aged between 35 and 65, 1460 responded to the initial letter but 500 did not wish to take part. 666 completed the questionnaire and of these 29 were assessed by the GP as being at high risk on the basis of criteria in use in local cancer genetics clinics. Around half of these people (14) had received genetic advice already. Among the remaining 15, who were referred to a cancer genetics clinic, one was found to have early signs of breast cancer. The paper does not report whether genetic testing was offered to any of the referred patients.

Comment: The results of the study indicate that screening for a high genetic risk of breast or colorectal cancer is possible within general practice. However the uptake was quite low (29%) and a significant percentage of patients (22%) actively declined to take part (rather than simply not responding).The authors suggest that uptake might have been higher if the invitation were made in another setting, such as in initial consultations with new patients. Looking beyond these specific points, however, from a public health perspective many important questions remain to be answered. What is the balance between benefit and harm in a screening programme such as this? (see item on population screening elsewhere in this Newsletter for a discussion of the twin problems of undue anxiety and false reassurance). What would be the costs of extending such a programme nationally, both in the administration and evaluation of the family-history questionnaire, and in the surveillance of those identified as being at high risk? Would GPs who do not have a personal interest in genetics be willing to take on such a programme? Would such screening divert patients' attention from the large environmental component of cancer risk? 

A storm of publicity surrounded the recent report in Nature that scientists had genetically engineered a strain of mice with enhanced ability to learn and remember [Tang, Y-P. et al. (1999) Nature 401, 63-69]. Perhaps the most startling aspect of the work was that it was apparently so simple: they just added to the mouse's genome an extra copy of a gene that encodes a component of a protein that sits in the membrane of cells of the brain. The function of the protein is concerned with enabling the brain to detect and strengthen associations between incoming signals, an essential part of both learning and memory. In a variety of behavioural tests, the engineered mice both learned faster and retained their learning longer than normal mice.

Comment: Several articles in the September 13 issue of Time magazine discuss in more detail both the science and the potential ethical questions it raises for humans. Learning, memory and the other facets of what we call intelligence have traditionally been thought to be much too complex to be amenable to enhancement by alteration of a single gene, and of course there is no way of knowing whether a similar experiment in humans (assuming it were both technically feasible and legal, neither of which applies at present) would work. Even if it did, the enhanced learning and memory might turn out to be accompanied by psychiatric problems or even a shorter lifespan. This issue is unlikely to be a pressing one for public health, at least in the foreseeable future, but it is nevertheless important for its impact on public attitudes to genetic technology. Public and media responses to the work in mice highlight deep worries about the possibilities for "tinkering" with the human genome, and all those concerned with the potential future impact of genetic technology ignore them at their peril. In the sphere of public health genetics, it is important to convey the message that the hope is that it might be possible to identify the genetic determinants of health and disease in individuals, with the idea of manipulating the environment, rather than genes, in order to prolong healthy life

In an editorial in the BMJ, Austoker discusses some of the problems with gaining true informed consent in population screening programmes [Austoker, J. (1999) BMJ 319, 722-723 (18 Sept issue)]. In particular, she stresses that screening has potential harms as well as benefits; people should be aware of these when deciding whether to participate, and the potential harms should not be "glossed over" in an attempt to maximise participation. The harms include the dangers of unnecessary, and sometimes potentially dangerous, investigation of false-positive cases, as well as anxiety and (for false negatives) false reassurance. The public is insufficiently aware of the limitations of screening, assuming that false positive and false negative results indicate negligence or incompetence on the part of the professionals carrying out the screening test, rather than being an inevitable aspect of any screening programme that tries to achieve an optimal balance of sensitivity and specificity. Austoker emphasises the vital importance of providing accurate and balanced information to enable informed choice and the need, in addition, for education of health professionals involved in screening programmes. In this context, it is interesting that, in a systematic review of "decision aids" for people facing choices about medical treatment or screening tests - O'Connor et al found that decision aids improved knowledge in women deciding whether to opt for BRCA1 gene testing, but had no effect on the percentage opting to take the test [O'Connor, A.M. et al. (1999) BMJ 319, 731-734 (18 Sept issue). It is important to stress, though, that BRCA1 gene testing is not a "screening test" and that in the study reported, the offer of testing was made in the context of careful counselling. For practical and financial reasons this is not always the case in population screening programmes.

Comment: All of the issues raised by Austoker apply equally to population genetic screening programmes and should be an essential part of deciding whether such programmes should be implemented.