Statistics from Altmetric.com
Curry et al have produced guidance on the provision of electron microscopy testing in virus laboratories.1 Such local guidance is necessary to make expensive and time consuming diagnostic services such as electron microscopy cost-effective and clinically useful.
In their report they state that “specimens for electron microscopy examination should be unformed faecal specimens (not formed stools...)”. They justify this approach by reference to “...20 years of experience have shown (unpublished results) that these are more productive.” Effectively they reiterate the commonly accepted assumption that viruses of gastroenteritis are more likely to be found in liquid than in solid faecal specimens.
Our findings do not support this dogma.2 We examined 2568 specimens by electron microscopy. A virus was demonstrated in 8.6% of liquid, 19.9% of semisolid, and 25.2% of solid specimens (χ2 for linear trend, p < 0.0001) This observation was valid for both adenovirus (2.4%, 5.0%, and 6.6%) and rotavirus (5.2%, 13.6%, and 16.6%). Curry et al cite our study but consider that their impression (unpublished results) of 20 years experience is better evidence than our data. This is contrary to an evidence based approach. Curry and his colleagues should present their findings comparing formed and unformed samples from patients with gastroenteritis for peer review to substantiate their conclusions.
Before our study our experience had not led us to question that the liquid specimens were associated with the highest yield. The rationale for initiating our study was to validate rather than disprove our approach of excluding solid specimens during times of pressure of use of the electron microscope.
Our findings clearly show that solid faecal specimens at the end of an episode of diarrhoea had a higher diagnostic yield than liquid specimens taken at the peak of symptoms. This pattern of results fits closely with the biphasic pattern of excretion described for rotaviruses in various animal models.3–5 This finding has important implications for those such as Curry et al and others establishing diagnostic algorithms for the investigation of viral gastroenteritis.
For diagnosis of small round structured viruses (SRSVs), it has become increasingly obvious that EM is not an adequately sensitive technique. Curry et al state that EM remains the front line test where SRSV is suspected. We would strongly disagree and consider nested reverse transcriptase polymerase chain reaction to be more appropriate. In our hands the sensitivity of nested PCR is 80% in specimens from SRSV outbreaks and this is now our front line test for all adult cases of gastroenteritis.
We welcome the comments of McCaughey et al, and agree that “experience” is no substitute for quantified scientific results that have been peer reviewed. However, the methodology used (grid floatation on a clarified 10% faecal suspension1) in their publication2 is not recommended within the PHLS as a suitable procedure for concentrating small round viruses, including SRSVs, from stool samples (although the method may be adequate for rotaviruses and adenoviruses). In their study only 18 of 2568 faecal specimens were positive for SRSVs. This detection rate of 0.7% must be compared to the SRSV detection rate by EM at Manchester PHL of around 10% when using either ultracentrifugation3 or the ammonium sulphate precipitation method.4 It is also worth pointing out that the laboratory from which the grid floatation method originated1 has recently sought a different method because of the relatively poor detection of SRSVs in faecal samples (personal communication to AC). The lack of sensitivity of the floatation method and the low magnification used by McCaughey et al to examine the grids (34 000) may explain the paucity of small round viruses detected in their study.
Despite our significant reservations about their methodology, McCaughey et al2 make the point that formed stool specimens contain more virus particles than either semiformed or liquid stool specimens. The important point we made is that the specimens should be collected in the acute stage of the illness for maximum EM sensitivity when using appropriate methodology. In addition, when investigating outbreaks, we are faced with limiting the workload and this requires selection of specimens most likely to contain virus. The reality of the situation is that the laboratory usually receives only a small amount of faecal sample from suspected cases of gastroenteritis, often without adequate clinical information. Our policy would be to discard those samples that are fully formed (hard pellet appearance). At least by selecting such unformed stool specimens we know that those individuals had symptoms. In practice, if the only specimens received have been produced more than 48 hours after onset of symptoms, then we now send a selection for PCR investigation.
There is no doubt that molecular diagnostic methods for SRSV detection are more sensitive than EM and also have a significant advantage for detection of virus RNA in specimens taken more than 48 hours after the onset of symptoms. In an ideal world, with unlimited resources and availability of skilled staffing, all specimens from outbreaks thought to involve SRSVs would be tested by appropriate molecular methods and results would be available within, at most, 24 hours. For sound economic and technical reasons, the PHLS funded and developed PCR detection of SRSVs in two centres (CPHL, Colindale, and Bristol PHL, in collaboration with Southampton University). This was to ensure that the limited resources available were placed in the hands of experienced molecular biologists so that development could occur quickly and that any problems encountered were rapidly resolved. This worked extremely well and both centres now offer a PCR based diagnostic service for SRSV detection. However, results are not as immediate as electron microscopy. The PHLS EM network, with all its limitations, will remain an important part of outbreak investigation in Wales and England for the immediate future.