Background
Whipple's disease is a paradigm of the evolution of infectious disease knowledge [
1]. The disease was first described in 1907 by Whipple [
2], who based it on anatomopathological lesions identified in a patient at autopsy. For many years, it was considered to be a metabolic disorder; however, in 1952, a bacterial origin became suspected when antibiotic treatment proved effective [
3]. The first molecular identification of the bacterium associated with Whipple's disease (
Tropheryma whipplei), as well as the first culture, created a new field [
4,
5].
T. whipplei has been identified in the saliva and stool specimens of healthy people [
1]. The well-known and classic form of Whipple's disease, which is characterised by periodic acid-Schiff (PAS)-stained bacilli in infected small-bowel macrophages, represents only one rare clinical form of the infection that can be caused by
T. whipplei. Indeed, the bacterium has also been involved in subacute or chronic infections without gut lesions, such as endocarditis, encephalitis, uveitis, adenopathy and pulmonary and osteoarticular infections [
1,
6‐
9]. The diagnosis and management are different for each of these infections [
1,
9,
10]. Finally, acute
T. whipplei infections, such as pneumonia [
11,
12], gastroenteritis [
13] and transit bacteraemia [
14], have recently been reported.
Neurologic forms are often the most serious manifestations of
T. whipplei infection, particularly with respect to relapse treatment failure [
15]. It has been noticed that relapses of classic Whipple's disease can be exclusively cerebral, without any peripheral manifestations [
1]. We recently diagnosed infection with
T. whipplei in a patient who presented with progressive dementia and recent-onset obesity and responded dramatically to antibiotic treatment. This observation encouraged us to document additional cases of
T. whipplei infections using brain biopsies and cerebrospinal fluid specimens.
Discussion
We describe a clinical phenomenon of patients with
T. whipplei who present with isolated brain involvement and respond dramatically to antibiotic treatment. Their condition was poor and worsening at the time of the diagnosis. As in the review of the literature, the antibiotics address the progressive cognitive impairment. In our index patient, antibiotic therapy not only improved his clinical status but also caused a reversal of obesity, both at the time of the initial diagnosis and upon relapse. These observations are reminiscent of those described in 1952 after the first successful use of an antibiotic (chloramphenicol) to treat Whipple's disease, which suggested that the disease had a bacterial origin [
3].
The amount of
T. whipplei DNA in the cerebrospinal fluid specimens of the patients described in our series was low, as demonstrated by our quantitative PCR assays; however, we are confident that these results are accurate, because we performed 2 PCR tests on 2 different samples to confirm the diagnosis. Our attempts to isolate a strain from the cerebrospinal fluid specimens failed, although we were able to isolate strains from other cerebrospinal fluids from patients with classic Whipple's disease with a good percentage of success (60%). Even in the 2 brain biopsies from our patients, only very low copies of
T. whipplei DNA were detected. Moreover, these 2 brain biopsies were PAS negative and
T. whipplei immunohistochemistry negative. It is also important to note that these data are different from those of patients in the literature with positive PAS staining of brain biopsies. Indeed, there is a link between the clinical picture of our patients and
T. whipplei, but we were unable to identify histological findings compatible with described
T. whipplei-associated diseases. However, it has also recently been reported that
T. whipplei DNA is highly frequent in the skin biopsies of patients with classic Whipple's disease, whereas PAS staining is significantly less sensitive in the same group [
55]. Furthermore, these bacteria are alive, as a positive culture from a skin specimen has been obtained. Overall, these data show that PAS staining, which was long considered the gold standard for diagnosing
T. whipplei infection, is frequently negative, despite the presence of
T. whipplei infection.
Overall, only 4 (18%) out of 18 patients reviewed here reported a history of arthralgia, despite its description as a cardinal symptom of classic Whipple's disease (73%) [
1]. Our study patients reported no chronic diarrhoea, whereas it is observed in 81% of patients with classical Whipple's disease and 40.5% of patients with a neurologic presentation of Whipple's disease [
1]. It is interesting to note that 3 of our 5 patients reported a history of unexplained acute diarrhoea prior to the onset of neurologic symptoms, which may correspond to primary infection with
T. whipplei, as described previously in young children [
13]. Weight loss was observed for none of our patients and only one from the literature versus 93% of patients with classic Whipple's disease and 47% of patients with a neurologic presentation of Whipple's disease [
1]. Even more interesting, 2 of our 5 patients developed obesity during the primary illness or during the relapse. Several viruses have been reported to contribute to weight gain in animals [
56]. In humans, one adenovirus has been also identified as a possible contributor to weight gain [
56]. The most likely mechanisms are a central effect on appetite and energy expenditure associated with hypothalamic infiltration by the pathogen. This clinical picture may be a particular form caused by
T. whipplei. It appears that this clinical entity resembles a localised cerebral relapse of classic Whipple's disease when bacterial multiplication is relatively controlled. Another interesting observation is the deficient humoural responses against
T. whipplei in patients with
T. whipplei encephalitis, as have been described in patients with Whipple's disease compared with asymptomatic carriers [
23].
This clinical entity seems rare, as only 0.85% of cerebrospinal fluid specimens of patients with unexplained dementia or encephalitis tested in our laboratory were positive. However, our technique may lack sensitivity in some cases, as the repeat PCR assays were unable to detect
T. whipplei when brain MRI-spectroscopy showed lesions compatible with reactive disease at the time of clinical relapse. Furthermore, the lack of information about antibiotic treatment status prior to cerebrospinal fluid sampling for most of the patients must also lead us to regard this prevalence cautiously, as antibiotic use might explain some negative PCR results. As seen in our patients, PAS staining of brain biopsy specimens with
T. whipplei also lacks sensitivity. Furthermore, this tests lacks specificity, as perivascular aggregates of foamy PAS-positive macrophages in a reactive gliosis setting can be seen in other macrophage-rich cerebral disorders, such as demyelinating diseases, cerebral infarction, and a host of infectious diseases [
57]. The results of PAS-staining of brain biopsy specimens can neither confirm nor exclude the diagnosis of
T. whipplei encephalitis. The development of cognitive impairments associated with ataxia or obesity should point the clinician to a diagnosis of
T. whipplei encephalitis.
PCR testing of cerebrospinal fluid specimens for
T. whipplei in neurological patients is clinically reasonable, as the disease is fatal without specific treatment. As is the case for all diagnoses performed using PCR, caution is necessary, and a rigorous strategy should be applied when performing and interpreting the analyses due to the risk of false-positive results [
58]. A carefully checked positive PCR in cerebrospinal fluid specimens is sufficient for diagnosis. It is important to remember the patient for whom a Whipple's disease diagnosis was ruled out by negative PAS staining, despite positive PCR, and then finally confirmed using both techniques at the patient's autopsy [
59].
Another question concerns the management of this presentation. Advances in knowledge, with
in vitro tests and full genome sequencing of
T. whipplei, have shown that the usual long-term treatment based on trimethoprim-sulfamethoxazole is a sulfonamide monotherapy and that an alternative may be doxycycline and hydroxychloroquine, an alkalinising agent [
1,
60,
61]. However, it is noteworthy to remember that trimethoprim-sulfamethoxazole has replaced cyclines, which had the reputation of not crossing the blood-brain barrier in adequate amounts, an issue that has not yet been confirmed for doxycycline [
10]. That is why we first used a combination of doxycycline, hydroxychloroquine and trimethoprim-sulfamethoxazole to manage
T. whipplei encephalitis. However, recent evidence suggests that sulfadiazine is preferable; its efficacy is comparable to that of the sulfomethoxazole, but a higher dose of sulfadiazine can be used to improve its ability to cross the blood-brain barrier [
10]. Antibiotics had a quick effect. However, 3 of our 5 patients presented with clinical relapse despite at least 18 months of therapy. When the treatment was reintroduced, a quick effect was again observed. As for classic Whipple's disease, the optimal treatment for
T. whipplei encephalitis has not yet been determined [
10]. Overall, the data emphasise the need for long-term treatment and lifelong follow-up.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
DR designed the study and analysed the data. FF, FN, CP, JCA, PC, HL and JP collected and analysed the data. FF and DR wrote the manuscript. All authors read and approved the final manuscript.