A Clear Solution for Dirty Water

Turning water into wine may be among the most venerable of miracles, but for Greg Allgood, the real miracle has been turning dirty water into drinkable water. He once wowed an audience in a Malawi village, where hundreds of inhabitants along with the country’s Minister of Health watched him transform a sample of the only local source of drinking water. “There were gasps of excitement when the water turned from this horrible, muddy dark color to crystal clear and safe,” he recalls. 
 
Allgood was demonstrating PUR™, a modest-looking packet of powder that quickly turns turbid, health-threatening water into the kind of liquid most of us would pay to drink out of a bottle. PUR was developed in the late 1990s by household products giant Procter & Gamble (P&G) and shares its name—but not its technology—with home tap water filters sold by that company in developed nations. Now PUR occupies a place at the forefront of P&G’s Children’s Safe Drinking Water Program, a philanthropic initiative that Allgood directs. 
 
Allgood spends about a third of his time in places like Malawi where people have limited or no access to treated, potable water sources. Worldwide, as many as 2 billion people drink water extracted from shallow wells or polluted lakes and rivers, with nothing like the municipal treatment systems that are taken for granted in most of North America and Europe. In the few developing locales where such infrastructure might exist—and indeed, even in the richest nations on the planet—this resource can be ruined suddenly by a natural disaster like a hurricane, earthquake, or tsunami, creating an immediate, desperate, and widespread need for safe drinking water.

Alpha1-antitrypsin (a]AT) is one of nine distinct plasma proteins which have been characterized as protease inhibitors (Harpel, 1983); it is a glycoprotein synthesized in the liver and it is responsible for about 90 per cent of the tryptic inhibitory capacity of human serum (Sharp, 1976;Kew et al., 1978;Morse, 1978;Chio & Oon, 1979). i1AT is under genetic control, and more than 30 codominant alleles at a single chromosomal locus have been identified (Chapuis-Cellier, 1975;Cox, 1978;Morse, 1978). Some of these alleles (particularly the alleles Z, I, S, and perhaps P and other less common ones) have been associated with serum cz1AT deficiency of varying degree (Chapuis-Cellier, 1975;Morse, 1978;Chapuis-Cellier & Arnaud, 1979) and with the presence in the liver of periodic acid-Schiff(PAS)-positive, diastase-resistant globules (Sharp, 1976;Morse, 1978;Palmer et al., 1980). There appears to be an intriguing link between a1AT and hepatocellular carcinoma (HCC) but the evidence is not clear cut (Sharp, 1976;Kew et al., 1978;Morse, 1978;Kelly et al., 1979). We have studied the association between e1AT and HCC in a large group of HCC cases and matched controls, using adequate laboratory procedures "blindly" (with respect to disease status), in a European population with an unusually high incidence of HCC (Trichopoulos et al., 1982).

Patients and methods
Two hundred and forty patients were studied; they were Caucasians, of Greek nationality and residence, hospitalized in one of eight large hospitals of Athens during a 15-month period in 1976 and 1977. Among them 80 had HCC (confirmed histologically in 47 cases and by diagnostic alpha-fetoprotein (aFP) values in the remaining 33 cases); for each of them two control patients, matched for age and sex, were selected from the same hospitals with diagnoses other than neoplasm or liver disease. Among HCC patients ("cases") 69 (86%) were males and the average age was 63 years; among comparison patients ("controls") the corresponding figures were 138 males (86%) and 62 years. All patients were interviewed, and blood samples were taken from each.
Hepatitis B serologic markers were determined in all of the sera by radioimmunoassay; samples of 39 cases (49%) and 12 controls (8%) were found to be positive for hepatitis B surface antigen (HBsAg) and were considered to have chronic infection. The two HCC groups (HBsAg positive and HBsAg negative) were of similar age and sex (mean age, 63 years in both series; proportion of males 85% and 88% respectively). Therefore, age and sex adjustments were not necessary for comparisons between controls and either or both of the two HCC groups (Trichopoulos et al., 1978). Serum levels and phenotypes of cx1AT were determined by radial immunodiffusion and C) The Macmillan Press Ltd., 1984 & D.
electrofocusing in acrylamide gel, respectively (Vesterberg, 1973;Chapuis-Cellier, 1975). All determinations were performed blindly in the Department of Clinical Biochemistry of the Hospital "Edouard Herriot" in Lyon, France. Table I shows the distributions of HCC cases and controls by HBsAg status (positive or negative) and xlAT phenotype pattern. The four distributions are very similar and the observed differences can easily be explained by chance. Thus, comparison of all HCC cases with all controls, with respect to the four most common alAT phenotypes (and a fifth group combining all the "other" less common ones) gives a x2 with 4 degrees of freedom of 2.56, corresponding to P>0.5. Furthermore, there is no evidence in the present data that phenotypes associated with axAT deficiency are overrepresented among HCC cases; on the contrary, two heterozygotes for the Z allele were included in the control series whereas none was found among the cases. Lastly, among the cases 6% were homozygous and 14% heterozygous for the M2 allele; among controls the corresponding figures were almost identical (7% and 11%, respectively) providing no support to the hypothesis that this allele is associated with a substantial increase of HCC risk. Table II shows mean values of serum a;AT (in mglOOml-') in HCC cases and controls by a1AT phenotype and, for HCC cases, by HBsAg status.

Results
The mean value of serum a1AT is higher in HCC cases than in hospital controls (by -40%) and the difference is highly significant (P < 10 -9). Although HCC cases of both subgroups have elevated alAT values the elevation is more marked among HBsAg-negative cases than among HBsAg-positive cases; the corresponding mean values are -50% and 30% higher than the mean value among controls and they differ significantly from each other (P<0.01) as well as from the mean value of controls (P<10-9 and p< 10-6 respectively). It is of interest that the pattern HCC (HBsAg-)> HCC(HBsAg+)>Controls is evident not only in the total but also within every single alAT phenotype (with only one marginal exception, in M1M2). There was no significant (or even similar) difference in the average level of serum a1AT between HBsAg positive and negative individuals in the control group; the mean values were 489 and 430mg 100ml-1 respectively, and the P value for a t-test of their difference is 0.30. Recent reports (Trichopoulos et al., 1980;Lam et al., 1982) indicated that tobacco smoking may cause HBsAgnegative HCC. Since tobacco smokers in general have elevated concentrations of serum a1AT, we have explored whether the very high values of a,AT in HCC (and particularly in HBsAg-negative HCC) could reflect a specific association between a,AT and tobacco-related HBsAg-negative HCC. There is some evidence in the present data that this may be so, but it is far from conclusive. Among patients with HBsAg-negative HCC, markedly elevated concentrations of serum x1AT (>600mg 100mlt1) were found in 21/31 current smokers (68%) but only in 5/10 non-smokers (50%); however, the  difference is neither statistically significant nor dose-dependent (Table III). We have also investigated whether, among HCC patients, there is a positive correlation between the serum concentrations of aFP (after logtransformation) and a1AT. There is no such evidence in the present data; among HBsAgpositive HCC patients the correlation coefficient is + 0.04, whereas among HBsAg-negative HCC patients the corresponding value is +0.05.

Discussion
The association between alAT and HCC has been studied from several points of view. Individuals who are homozygous for the Z allele (and perhaps other alleles associated with a1AT deficiency) have an increased risk for HCC, and many of them have the characteristic globular bodies in their livers. However, it is not clear whether individuals who B are heterozygous for these alleles are at increased risk for HCC; many studies have noted modest associations but several others have not, even though heterozygotes have consistently the same characteristic globules in their livers (studies reviewed by Sharp, 1976;Kew et al., 1978;Morse, 1978;Kelly et al., 1979;Sizaret et al., 1981;Spech & Liehr, 1982). Kew et al. (1978) have summarized the evidence by stating that a1AT deficiency could be an occassional cause of the tumour in most parts of the world where HCC occurs sporadically, but it is not a numerically important cause in Africa and the Far East where HCC is common. Our findings support this conclusion and permit its generalization to other populations where HCC is common, besides those of Africa and Asia. On the other hand, we did not confirm the associations of HCC with the F or the M2 alleles, which were reported by Theodoropoulos et al. Sizaret et al. (1981), respectively. Since the present study is larger than the other two we are inclined to believe that the earlier findings were, perhaps, fortuitous.
Elevated values of serum a,AT in HCC cases have been noted by Kew et al., (1978), Chio &Oon (1979), andMatsuzaki et al. (1981). Our findings confirm the results of the earlier investigations and indicate that the elevation is sufficiently marked to be aetiologically intriguing and diagnostically useful. Furthermore, we have found that serum a1AT values are, on the average, higher in HBsAgnegative cases of HCC than in HBsAg-positive cases of this tumour-an observation not reported previously. Although the difference is not sufficiently large to be of clinical importance it suggests that the aetiologic heterogeneity of HCC is reflected not only in the epidemiologic parameters but also in the laboratory findings of the aetiologic subgroups.
The molecular, biochemical and histological aspects of the association between aeAT and HCC have been reviewed elsewhere (Sharp, 1976;Morse, 1978;Palmer et al., 1980;Spech & Liehr, 1982). However, the differential increase of acAT in HBsAg-positive and HBsAg-negative HCC, if real, calls for explanation. Several possibilities exist. Tobacco smoking has been found to increase the serum levels of alAT by -20% (Lellouch et al., 1979) and this proportional excess may also concern the tobacco related HCC cases. It should be noted, also, that cirrhosis is associated with reduced levels of serum alAT (Chio & Oon, 1979;Matsuzaki et al., 1981), and cirrhosis (in Greece, at least) is more frequently associated with HBsAgpositive than with HBsAg-negative HCC (Trichopoulos et al., 1982). Palmer et al. (1980), using ultrastructural, histochemical and immunocytochemical methods have demonstrated the occasional parallel emergence of aFP and acAT as tumour tissue markers in malignant hepatoma (HCC). It is, therefore, surprising that a significant positive association was not evident between the serum values of aFP and a1AT in the present study or in the earlier study of Chio & Oon (1979). Although chance is a likely explanation for the absence of a significant association, it is also possible that the epigenetic emergence of ajAT as tumour tissue marker is not frequently accompanied by an increase in the serum levels of ciAT (Palmer et al., 1980). Supported by a grant from the Greek Ministry of Health.