Tropheryma whipplei in the Environment: Survey of Sewage Plant Influxes and Sewage Plant Workers

Whipple's disease is a rare multisystem inflammatory disease, with chronic diarrhea, weight loss, ascites, and arthralgia as the most prominent symptoms (16), caused by Tropheryma whipplei, a member of the family of actinomycetes (20, 23). The diagnostic hallmark is gram-positive bacteria in the duodenal mucosa, which stain positively with periodic acid-Schiff stain; diagnosis is now usually confirmed by amplifying T. whipplei-specific DNA sequences (2, 17, 18). As little is known about the natural habitat of T. whipplei, we investigated 46 influxes to 14 sewage treatment plants in the vicinity of Vienna, Austria, for the presence of T. whipplei by PCR (17). Although workers at those plants are not known to have a higher prevalence of Whipple's disease, it is reasonable to assume that they are at increased risk for contact to T. whipplei. Therefore, we screened 64 healthy workers (all male; mean age, 42 years; range, 25 to 65 years) at those sewage treatment plants for the presence of that bacterium in stool (9), dental plaques (4), and mouthwash fluid (22) and compared the results with those obtained from 146 disease controls (88 males and 58 females; mean age, 55 years; range, 29 to 82 years), recruited from a gastroenterology ward, without known increased risk for contact with T. whipplei in the environment. Furthermore, we tried to quantitate the number of T. whipplei genomes in sewage treatment plant influxes and T. whipplei-positive stool samples.

DNA samples from sewage plant influxes, mouthwash fluids, and dental plaques were prepared using standard methods; DNA from stool samples was recovered using a spin stool DNA kit (Invitek, Germany).

PCR amplification of T. whipplei DNA was performed as described previously (17), and sensitivity, corresponding to one to three genomes of T. whipplei in the PCR and 5 to 10 genomes/ml sample material, was estimated using a DNA sample from cultured T. whipplei (kindly provided by B. LaScola, Marseille, France). Specificity of positive samples was confirmed by direct sequencing (13, 17). For quantitation of T. whipplei genomes, a real-time PCR using the primers TW13F and TW163R (sequences kindly provided by B. LaScola, Marseille) was established, with a sensitivity of 5,000 genomes/ml sample volume (5).

T. whipplei was found by PCR in 17 of 46 (37%) influxes to sewage treatment plants (Table 1) . The 17 influxes testing positive came from 47% of the residential and agricultural communities, whereas all influxes from industrial sites remained negative (P = 0.008). This percentage of T. whipplei-positive wastewater samples is similar to but slightly lower than that determined by others (15). This could be explained by differences in the agricultural or industrial mixes of the communities served by the sewage treatment plants. Another possibility is a difference in the sensitivities of the detection methods. Sensitivity of our nested PCR was 5 to 10 genomes/ml, which is similar to that reached by others using PCR methods (7). Real-time PCR proved to be somewhat less sensitive (5,000 genomes/ml), but this represents the lower threshold of the linear measuring range, below which no reliable quantitative information can be given. By using both nested PCR and real-time PCR, we were able to estimate the concentration of T. whipplei genomes in samples of sewage plant influxes to be between 5 to 10 and 5,000 genomes/ml sample volume. This is a very low number of bacteria in sewage plant influxes compared to concentrations of Escherichia coli (>10⁵ CFU/ml), Enterococcus spp. (>10⁴/ml), or other indicator bacteria (10). In addition, culture-independent enumeration methods usually give numbers 2 log higher than traditional culture-based detection systems (12). This low number compared to other bacteria is well compatible with the already known prolonged replication time of approximately 18 days of T. whipplei in fibroblasts in vitro (19, 21). Furthermore, this explains why this bacterium is difficult to find in the polymicrobial environment of wastewater or other environmental sources (15).

Sixteen (25%) of the 64 sewage plant workers tested positive for T. whipplei in stool by nested PCR (Table 1), whereas T. whipplei was found in stools of only 12 of 178 (7%) disease controls (without antibiotic treatment at this point in time [6]) (P = 0.001; Fisher's exact test). Amsler (1) reported the occurrence of T. whipplei in stool in 5 of 196 (2.5%) disease control patients, which is lower than our figure of 7%. The same group reported 13 of 208 stool samples to be positive for T. whipplei (6%) in healthy control individuals, which is similar to our disease control group but considerably less than the number of positive stool samples found in workers at sewage treatment plants. Our findings are compatible with the notion that close contact to sewage water increases the likelihood of excreting T. whipplei in stool.

In disease control patients, T. whipplei-positive stool samples were found more often in patients with liver cirrhosis (12.5%) than in patients with other diseases (2%) (P = 0.014). In patients with liver cirrhosis, 9 of the 10 positive stool samples were found in patients who had developed hepatocellular carcinoma in the cirrhotic liver (14%), and one positive stool sample was from a patient without hepatocellular carcinoma (5%) (P = 0.473). This high percentage of excretors of T. whipplei in patients with liver cirrhosis and hepatocellular carcinoma remains unexplained. In patients without liver disease, one positive sample was found in a patient with chronic obstructive pulmonary disease and one positive stool sample in a patient with chronic cardiac disease, whereas all other patients with various diagnoses were negative for T. whipplei in stool. It is noteworthy that neither in the groups of patients with immunosuppression (a human immunodeficiency virus-positive patient and patients with kidney, heart, or liver transplants) nor in the groups of patients immunocompromised due to malignant diseases (leukemia, non-Hodgkin's lymphoma, myelodysplastic syndrome, or solid tumors) was any stool sample found to be positive for T. whipplei. This finding argues for a more specific immune derangement necessary to result in colonization of the colon with T. whipplei. Indeed, a T. whipplei-specific reduced Th1 response has been described recently for patients with Whipple's disease (3, 16a).

In 51 of the 64 sewage plant workers, a second stool sample could be investigated. Five of the 16 (31%) workers with a stool positive for T. whipplei had this bacterium also in a second stool sample. Repeat stool samples could be obtained from 5 of the 12 disease controls with T. whipplei in stool; in 4 of the 5, the bacterium was found again. In three patients (all suffering from liver cirrhosis and hepatocellular carcinoma), a third and a fourth stool sample taken at monthly intervals were also positive. In the last three disease controls, a gastroscopy with duodenal biopsy was performed. PCR from duodenal mucosa was positive in one sample and negative in two others, and histologic staining with periodic acid-Schiff stain was negative in all three disease controls.

The repeat finding of T. whipplei in stool of both workers and disease control patients argued that the positive finding of T. whipplei in stool is not the result of a false-positive test but that an excretor status is a consistent phenomenon in some individuals.

In 2 of the 28 positive stool samples from sewage plant workers and disease control patients, T. whipplei genomes could be quantitated in stool by real-time PCR. One stool sample contained 5,333 genomes/ml stool, and the other contained 32,000 genomes/ml. Twenty-six PCR-positive stool samples were below the sensitivity limit of the real-time PCR. The number of T. whipplei genomes in those stool samples was estimated to be between 5 to 10 and 5,000 genomes/ml stool. These numbers are very low for bacterial counts/ml stool compared to other bacteria known to be present in abundant numbers in stool. For instance, E. coli usually has a count of 10⁶ to 10⁷ CFU/ml stool (11), Bifidobacterium spp. of >10⁹ CFU/ml (14), and Bacteroides spp. of >10¹⁰ cells/ml (8), with much higher actual numbers of bacteria obtained by culture-independent methods (12).

DNA from T. whipplei was not found by PCR in any of the dental plaque samples, similar to the results of Dutly et al. (4), but was found in the mouthwash fluid of 1 of 64 (1.5%) sewage plant workers. This worker was also positive for T. whipplei in stool. When related to those workers with T. whipplei-positive stools, this corresponds to 6.25%. Our results are lower than the numbers of positive findings in a paper by Street et al. (22) and by Dutly et al. (4), who reported the presence of T. whipplei in saliva in 6 of 14 individuals in whom gastric juice was found positive for T. whipplei by PCR (42%). Amsler et al. (1) reported a prevalence of T. whipplei in saliva in 3 of 215 healthy controls, which corresponds to a positivity rate of 1.3%, similar to our findings of 1.5%.

• American Society for Microbiology