Microsatellite Typing of Cryptosporidium parvum in Isolates from a Waterborne Outbreak

A recent study reported microsatellite subtyping of isolates from cases in five drinking water outbreaks (3) but lacked patient specific exposure data, which prevented the demonstration of the full value of subtyping in outbreak investigation. We report the use of multilocus microsatellite subtyping to investigate an outbreak of drinking-water-associated cryptosporidiosis in the northwest of England which affected residents during April and May 2000 and in which exposure data were available.

The outbreak was in an area where there hade been multiple outbreaks over previous years (4). Some 238 cases were identified in the affected areas, though not all were thought to be related to the primary outbreak. The typing methods used have been extensively described in our previous papers (1, 2). A total of 99 strains were typed, of which 10 strains of Cryptosporidium hominis and 78 of Cryptosporidium parvum were typeable at all three loci. The distribution of multilocus fragment types (MLFTs) for these typeable strains is shown in Table 1. For C. parvum there were 17 different MLFTs detected; the most common type (P36) accounted for 37% (29) of strains, and the second (P5) accounted for 19% of strains. Ten of the detected types (P34 to P45) were distinct from strain types identified in the previously reported study of sporadic infections (2). All but four of the C. parvum strains had the ML1 242 allele previously shown to be associated with zoonotic infection (2).

TABLE 1.

Distribution of Cryptosporidium hominis and Cryptosporidium parvum MLFTs

Cryptosporidium species (n)	MLFT	No. of cases	Allele no.
			ML1	ML2	gp60
C. hominis (10)	H1	10	233	180	371
C. parvum (78)	P1	6	242	229	341
	P2	2	242	229	338
	P5	15	242	231	341
	P7	1	242	231	338
	P8	4	242	233	341
	P14	6	242	231	356
	P19	1	242	235	338
	P34	5	242	207	338
	P36	29	242	229	344
	P37	1	242	211	344
	P38	1	227	197	368
	P39	2	242	231	362
	P40	1	242	221	341
	P41	1	242	235	341
	P42	1	245	229	338
	P44	1	227	197	335
	P45	1	239	231	341

Illness associated with C. parvum reached a peak in week 4, and both P36 and P5 had the greatest number of associated cases in that week. There was no significant difference between the number of weekly cases of the most common MLFT (P36) and those of all other C. parvum MLFTs (Wilcoxon signed-rank test; Z = −1.552, P = 0.121).

Figure 1 shows the geographical distribution of the two main strains and other C. parvum strains. Nearly all P36 strains were strongly clustered in the area around Preston and Chorley, the main focus for the outbreak and the area most dependent on water from the implicated supply. When comparing the distances of each of the P36 strains to their geographic centroid with that for other strains of C. parvum, this clustering of P36 was highly significant (Mann-Whitney U test, Z = 4.580, P = 0.000004).

FIG. 1.

Geographical distribution of P36, P5, and other C. parvum infection strains.

The value of typing in this outbreak was the demonstration that most strains were of the zoonotic ML1 242 allele. Subtyping enabled identification of a particularly common type but one that represented only 37% of strains. Subtyping also demonstrated the lack of this type in areas not reliant on the affected water supply. It currently remains unclear whether the variation in subtypes identified represented different lineages or evolution of strains during the outbreak. The issue of the stability of MLFTs in natural infections needs to be investigated in more detail. In this context it is worth noting that a further nine strains differed from P36 at only one locus and so could potentially be derived from P36 with just a single mutation. Microsatellite typing of Cryptosporidium isolates associated with waterborne outbreaks can provide useful additional epidemiological information especially when linked to individual exposure data.