Part A. Systematic qualitative review

Method

MEDLINE (via Ebscohost) and PubMed databases were searched for publications from 1965 to April 1 2016. To obtain papers referring to d-la, the following search terms were used: d-lact* AND (acidosis OR encephalopathy OR neuropathy). These databases were also searched for ME/CFS articles referring to acidosis (search terms: acidosis AND (“chronic fatigue syndrome” OR “myalgic encephalomyelitis”). Reference lists from articles obtained were manually screened to find other relevant references. Figure 1 shows the PRISMA flowchart for identification, screening and article exclusion.

Systematic summary of d-la episodes

Table 3 shows some evidence of shared symptomatology across each broad ME/CFS ICC category. The highest percentage of overlap was found for neurological symptoms. Other symptoms specific to d-la were also frequently reported (e.g., metabolic acidosis). ME/CFS symptom categories are discussed sequentially to examine similarities with d-la symptoms.

Overlap between d-la and ME/CFS symptoms

Neurological impairments

Episodes reviewed in this qualitative synthesis required neurological symptoms to be reported (as an inclusion criterion), accordingly, all episodes of d-la reported at least one neurological impairment. The majority of neurological symptoms that were reported overlapped with ME/CFS symptomatology (52/59, 88.1%). The frequencies of matching ME/CFS neurological symptoms were similar when comparing paediatric (19/22, 86.4%) and adult (33/37, 89.2%) episodes as well as male (30/35, 85.7%) and female (21/23, 91.3%) episodes. Ambiguous/other neurological impairments (e.g., altered mental state or cortical dysfunction) were also frequently reported (45/59, 76.3%). The more severe neurological symptom of an altered level of consciousness was reported in 13 episodes (22.0%). Five case reports documented the patient’s altered consciousness as the only neurological symptom during the d-la episode. The remaining reports described additional neurological symptoms and a deterioration in symptoms affecting consciousness.

When considering more specific types of neurological impairments, motor disturbance (B4b) was the most frequently reported matching ME/CFS neurological symptom (42/59, 71.2%, see Fig. 2). This was notably higher than the other neurological symptoms (B1. Neurocognitive = 25/59, 42.4%, B2. Pain = 3/59, 5.1%, B3. Sleep = 10/59, 16.9%, B4a. Neurosensory and Perceptual = 2/59, 3.4%). Common motor disturbances in ME/CFS include muscle weakness, clumsiness, balance and coordination difficulties [84]. The ICC noted that the presence of balance and gait instabilities are more frequently observed in severe cases [1]. Ambiguous/other neurocognitive, neurosensory, perceptual and motor disturbances were reported in 37.3% of total d-la episodes (22/59). Within these 22 episodes, 90.9% (20/22) of episodes simultaneously reported at least one matching neurological symptom akin with ME/CFS diagnostic criteria. Therefore, there was considerable overlap between matching symptoms and ambiguous/other neurological symptoms.

Approximately half of d-la episodes reported impairments in speech (30/59, 50.9%). Notably, all episodes that reported speech and language impairments also reported at least one other ME/CFS-matching neurological impairment, which may reflect the shared pathophysiology that underlies the behavioural manifestation of overt speech symptoms.

Limitations

These results need to be considered with an awareness of potential methodological limitations. Firstly, the inclusion criteria for selected case reports in this review may have been problematic. Although unavoidable, the requirement of neurological symptoms during d-la episodes may have increased reporting bias during this review process leading to an exaggerated focus on neurological symptoms. However, the effect of this limitation may be moderated when considering the high percentage of case reports meeting both the eligibility criteria of describing neurological symptoms and d-lactate measurement during the episode (80.0%).

Findings from this qualitative review are also limited by the lack of standardised procedures used when reporting symptoms in case reports. Differences in assessment procedures and terminology used for reporting neurological symptoms may impede accurate interpretation. Some ambiguous/other symptoms described as distinct may share similar pathophysiology. This may be particularly pertinent for speech symptoms. On the one hand, the results may underestimate the level of overlap based on the conservative classification of symptoms. Alternatively, the breadth of ME/CFS symptoms included in the ICC diagnostic criteria may inflate the findings. Reliance on qualitative symptom report comparisons only provide a preliminary guide to shared symptomatology. Whilst useful for theoretical purposes it is insufficient to draw confirmatory conclusions.

Implications

Mindful of these limitations, the proposal of a continuum of acute and chronic encephalopathy related to d-lactate warrants further investigation. Within d-la, several authors have proposed that the level of acidosis and associated encephalopathy may result in differing severity and either an acute or chronic presentation [27, 28, 32]. A subclinical elevation of d-lactate has been reported in SBS patients and diverse populations [94]. Higher d-lactate levels were recorded in 2.8% of randomly selected hospital patients [40]. Minimal details were provided about this sample other than noting that 40% of these patients did not have a history of gastrointestinal surgery [40]. Higher levels of d-lactate have also been recorded in response to trauma or infection [95]; and in patients with diabetes compared with healthy controls [96]. Thornalley et al. [97] showed positive correlations between the level of d-lactate and duration of diabetes. They found that the duration of disease was positively associated with retinopathy, neuropathy and nephropathy complications of diabetes. The relevance of d-lactate for diverse presentations is currently unknown.

Even within SBS populations, d-la has been under-recognised and frequently misdiagnosed [9]. Misdiagnosis is complicated by issues related to accurate and efficient measurement of d-lactate. A further diagnostic complication related to the clinical presentation of d-la is that the neurological manifestations can present without gastrointestinal complications or a change in gastrointestinal symptoms. Less than one-quarter of d-la episodes analysed in this review described a worsening of gastrointestinal symptoms. Therefore, it is plausible that clinicians may focus on the neurological presentation and overlook the underlying gastrointestinal mechanism. The case report from Scully et al. [70] highlighted this when the 16-year old male patient was first treated by a psychiatrist with lithium carbonate for suspected bipolar disorder and tested for illicit drug use before being diagnosed with d-la. The patient presented with aggression, somnolence and weight loss without current gastrointestinal symptoms, although had an abdominal trauma one year prior that required short-bowel surgery [70]. The presence of neurological symptoms in the absence of current gastrointestinal symptoms may lead to frequent misdiagnoses. The proposed mechanisms of d-la (i.e., carbohydrate malabsorption and related bacterial overgrowth [32, 36]) may have relevance for patients presenting with neurological symptoms but without an observable change in gastrointestinal symptoms.

Carbohydrate malabsorption is not exclusive to SBS populations and can vary in severity. Altschule et al. [98] found that d-lactate was more slowly metabolised in patients with schizophrenia, manic-depression and psychosis compared with healthy controls. Even earlier studies have shown increased lactate after fructose or glucose ingestion and disturbed lactate metabolism after exercise within these populations [98], suggesting difficulties with carbohydrate metabolism. Within ME/CFS, carbohydrate restriction (e.g., avoidance of sugars and grains) may be advantageous [99, 100]. Whilst there is minimal empirical support, clinical reports suggest that dietary triggers can exacerbate symptoms and that some patients benefit from dietary exclusions [99]. The response to treatment for small intestine bacterial overgrowth (SIBO) in ME/CFS patients [101], suggests that carbohydrate malabsorption may be relevant for a subgroup of this population. An analysis of mechanisms involved in d-la is provided to help identify shared pathophysiology between d-la and ME/CFS.

Proposed mechanisms in d-lactic acidosis

Understanding d-la involves firstly establishing the reason for increased d-lactate levels before examining proposed neurological mechanisms. Figure 4 summarises the contextual factors, triggers and proposed mechanisms leading to d-la. The presentation of d-la requires both an increase in d-lactate absorption that exceeds the metabolic and/or excretion capacity of the patient.

Proposed neurological mechanisms in d-lactic acidosis

Metabolic acidosis and increased d-lactate levels are synonymous with the presentation of d-la. However, neither condition can predict the development of neurological symptoms. Acidosis can occur without associated neurological symptoms and in reverse, encephalopathy can be present without the accompanying acidosis [13]. Similarly, whilst increased d-lactate levels are required in d-la, the presence of high d-lactate is not the sole cause or determinant of neurological symptoms. Some studies have shown a temporal association between d-lactate level and symptom onset [30] as well as severity [103]. However, this has not always been replicated (e.g., [48]). Other factors must also be required because higher d-lactate levels have been shown in patients with SBS and other gastrointestinal disorders but without concurrent encephalopathy [111]. These inconsistencies suggest that there are several possible direct and indirect mechanisms responsible for the neurological manifestations in d-la (see Fig. 5).

Summary

There is more support for the indirect effect of d-lactate interfering with energy metabolism in the CNS compared with the direct neurotoxic effects of d-lactate. Multiple mechanisms may be at play. Evidence of the inhibitory action of d-lactate on utilisation of l-lactate in neural cells and astroglia appears a particularly pertinent mechanism that may explain the neurological symptoms observed in d-la. The relevance of other bacterial metabolites remains in question. The vulnerability of certain individuals related to predisposing genetic, microbial factors or nutritional status that influence d-lactate production and/or adequate excretion/metabolism is likely to contribute to the presentation of d-la.

What is the relevance for ME/CFS?

Whilst d-lactate levels have not been specifically measured in ME/CFS patients, elevated lactate levels within ventricular cerebrospinal fluid have been observed. Significantly higher levels of ventricular lactate were recorded in the ME/CFS patient group compared to both generalized anxiety disorder (GAD) and control groups. From this small sample of 16 CFS patients, 10 patients had high ventricular lactate levels, indicated by lactate levels above 2 standard deviations above control mean whereas the remaining 6 participants had equivalent lactate levels to both the GAD and healthy control groups. This distinction between clinical and control groups gives promise for ventricular lactate being a potential biomarker useful for establishing ME/CFS subgroups. Interestingly lactate level was not associated with any other demographic or clinical variables, including severity of illness. Notably, clinical measures of anxiety, depression, fatigue, sleep quality and fibromyalgia were used as outcome variables but cognitive symptoms were not measured. More detailed analysis of associations between objective neurocognitive symptoms and ventricular lactate level would be valuable. The authors explained the potential mechanisms related to mitochondrial dysfunction and/or oxidative stress that precede reduced cerebral blood flow which in turn upregulates anaerobic glycolysis and consequential lactate accumulation [131]. Mitochondrial dysfunction or increased oxidative stress may have bacterial and/or viral origins, or be related to underlying gastrointestinal abnormalities.

Gastrointestinal abnormalities

Examination of gastrointestinal abnormalities in ME/CFS indicate some similarities between d-la mechanisms and ME/CFS pathophysiology. Gastrointestinal dysfunction is included as one of the multiple symptoms in ME/CFS. Although not required for a diagnosis, gastrointestinal abnormalities and comorbid irritable bowel syndrome are frequently reported by patients with ME/CFS [132]. ME/CFS patients more frequently experience gastrointestinal symptoms and use corresponding treatments (i.e., antacids, H2 blockers, proton pump inhibitors) compared with healthy controls [133]. Estimates based on a clinical patient group of 1400 patients show recurring gastrointestinal symptoms are experienced by between 80 and 90% of patients [134]. In a sample of 165 CFS patients, Chia and Chia [134] identified evidence of chronic inflammation and enterovirus of the stomach in 95 and 82% of patient biopsies respectively. As the authors suggest, the presence of viral markers in the stomach years after initial infection suggest that chronic viral infections of the stomach may contribute to continued pathophysiology. Viral infections have been proposed to precipitate and perpetuate the bacterial dysbiosis observed in ME/CFS (see review by Navaneetharaja et al. [135]).

Bacterial dysbiosis, antibiotic and probiotic treatment

Evidence of gut dysbiosis has been observed through measurement of fecal microbial composition in ME/CFS populations. Differences between microbial composition of healthy compared with ME/CFS populations have been reported using both culture-based [136, 137] and genetic sequencing methods [3, 138]. Treatment using antibiotic [139], probiotic [140–142] or bacteriotherapy [143] have also been used to help modulate the gut microbiota in ME/CFS with somewhat unpredictable and varied success.

Using culture-based methods, we have previously observed a predominance of d-lactate producing bacteria (Enterococcus and Streptococcus species) in ME/CFS patients [4]. These bacteria produce high levels of lactate in vitro, compared with fecal isolates [4] which would support the maintenance of a more acidic colonic environment as one of the mechanisms in d-la that was previously described. This inference about the acidity of the colon in ME/CFS patients has been deduced from in vitro methods only, as we are not aware of any research that has measured colonic pH in this population. Within our prior clinical investigations, responders to a short-term antibiotic treatment for Streptococcus overgrowth was associated with increased vigor on self-reports and selected improvement on objective sleep markers [139]. Extending from these preliminary findings, we are currently examining interactions between microbiota, broader neuropsychological symptoms and d-lactate levels in a clinical pilot evaluating treatment aimed at reducing an overgrowth of Streptococcus in a subgroup of ME/CFS patients. Our group have also compared culture-based fecal assessment and symptom expression within a larger sample (N = 274) of ME/CFS patients [5]. This observational study showed partial support for d-lactate theory in ME/CFS whilst raising questions about sex differences. Significant positive associations between some lactate producing bacteria (Lactobacillus and Streptococcus genera) and ME/CFS symptoms were shown for males but not females [5, 144]. Notably, the relative abundance of genera measured was consistent across the sexes raising questions about the functional differences of microbiota or a differing response to d-lactate for males compared to females. The heterogeneity of presentation and differing response to treatments could have varied explanations. Through the d-lactate lens, a preferential uptake of d-lactate (i.e., d-lactate accumulation as proposed by Mayeur et al. [110]) may contribute to variable symptoms and/or treatment response.

Using sequencing methods, Frémont et al. [3] examined ME/CFS patients and healthy controls from Norway and Belgium. Comparison between patient and control groups revealed no significant difference in bacterial diversity across the groups but differences in composition were observed. When comparing Norwegian patient and control samples there was a significant difference in bacterial composition, with ME/CFS patients showing a lower proportion of genus within the Firmicutes phylum. Interestingly, microbial differences between culturally diverse control samples (i.e., Norwegian compared with Belgian; [3] highlight the importance of considering inter-individual characteristics that may contribute to microbial variation.

Unlike Frémont et al. [3] findings of similar bacterial diversity, Giloteaux et al. [138] reported evidence of decreased diversity of microbial composition and instability in the microbial community in ME/CFS patients compared with controls. Significant differences were not shown when comparing the composition of ME/CFS and control samples at the phylum or genus level. However, at the operational taxonomic unit (OTU) level, proportions significantly differed for 40 OTU’s. For example, the proportion of Faecalibacterium and Bifidobacterium was significantly lower in ME/CFS patients compared with controls. The few studies that have examined fecal microbial composition in ME/CFS have shown some inconsistent results making current interpretation incomplete suggesting that evaluation of subgroups, species-level comparison and measurement of metabolites is required. Replication using a combination of culturing and genetic sequencing methods with larger samples and varied demographics will help ascertain the relevance of d-lactate neurotoxicity in ME/CFS.

Bifidobacterium are high lactate-producing bacteria. Whilst the ratio of D/L lactate vary between species, a lower proportion of Bifidobacterium species raises some doubt about the relevance of d-lactate theory for ME/CFS. Selected Bifidobacterium (Bifidobacterium adolescentis, Bifidobacterium breve) and Lactobacillus (L. plantarum, L. salivarius, L. casei subspecies rhamnosus, L. delbrueckii subsp. Lactis, L. acidophilus, L. fermentum, L. buchneri) species have been identified as predominant in patients with d-la [26, 28, 29, 37, 47, 54, 64, 68, 74, 75, 145]. Similarity between species identified as overgrown in d-la patients and those used in probiotic studies could also generate skepticism about the relevance of d-lactate theory for ME/CFS. Both a small open-label [140] and two randomized, double-blind placebo-controlled studies [141, 142] examining the efficacy of probiotic supplementation in ME/CFS have indicated modest improvements for selected symptoms.

ME/CFS patients supplemented with a lactic-acid producing bacterial strain probiotic showed clinical improvement in self-reported neurological symptoms but no significant changes in fatigue or activity levels [140]. Rao et al.’s (2009) small double-blind RCT used an eight-week probiotic supplementation of Lactobacillus casei to examine changes in emotional symptoms. ME/CFS patients in the treatment group reported a significant reduction in anxiety symptoms compared with controls. No change was recorded for subjective reports of depression. More recently, treatment using Bifidobacterium infantis 35,624 resulted in reduced inflammation in ME/CFS patients, however neurological symptoms were not measured [142]. Preliminary results indicate the need for further investigation of the efficacy of probiotic treatment in ME/CFS. Of relevance to the current hypothesis in question, the d-lactate potential of selected strains used in the aforementioned studies is unknown. Therefore, results from these treatment studies suggest support for gut–brain interaction in ME/CFS but fail to provide additional information about the relevance of d-lactate for this population.

Bacterial overgrowth in the small intestine may also have implications for d-lactate production. Logan et al. [146] hypothesized that SIBO is involved in ME/CFS and related to the immune alterations observed in this condition. SIBO can be a cause of functional short bowel and result in carbohydrate malabsorption. Patients with comorbid SIBO and CFS have shown clinical improvement (on subjective reports of depression, memory/concentration and pain) following antibiotic treatment [101]. d-Lactate levels were not measured in this study but dependent on the type of bacterial overgrowth, excess production of bacterial metabolites (including but not limited to d-lactate) may act centrally, through ENS activation or systemically due to intestinal permeability.

Implications for gut–brain interaction

Systemic inflammation as a consequence of gut mucosal damage and intestinal permeability as the first proposed pathway of gut–brain interaction in d-la has also been suggested as a pathophysiological mechanism in ME/CFS [116]. Initial support for this hypothesis in ME/CFS is reflected by findings of an increased immune response to lipopolysaccharide (LPS) (as measured by serum IgA and IgM to selected bacteria [116] and clinical improvement after treatment to restore intestinal permeability [113]. Measurement of plasma levels of LPS have been used as an indicator of microbial translocation as they are produced in response to Gram-negative bacteria [138]. Chronic LPS stimulation can be measured by plasma sCD14 and plasma LPS-binding protein (LBP) levels [138]. Recently, additional evidence of intestinal permeability in ME/CFS patients has been shown through significantly higher proportions of plasma LPS, LBP and sCD14 compared with controls [138]. These results support the hypothesis of an inflammatory and/or immune response to microbial translocation that occurs when there is chronic gut mucosal damage and intestinal permeability in ME/CFS patients.

Nutritional deficiencies in ME/CFS

Nutritional status can be impaired for individuals with chronic health conditions and comorbid gastrointestinal abnormalities. Nutritional deficiencies require careful monitoring and treatment for ME/CFS patients [1]. Coenzyme Q10 (CoQ10) was shown to be significantly lower in the plasma of ME/CFS patients compared with healthy controls [147]. Treatment that includes nutritional supplementation is frequently employed with CoQ10, magnesium, l-carnitine and S-adenosylmethionine indicated as potentially beneficial for this population [148, 149]. Improvements in cognitive symptoms (mental fatigue, attention, concentration) have been described after supplementation with acetyl-L-carnitine and propionyl-L-carnitine for patients with ME/CFS [150]. Colabamin (B12) injections are proposed to exert effects by reducing oxidative stress [151] but the implications of B12 deficiency may also be relevant when considering the role of B12 in the TCA cycle and lactate metabolism (see [152]). Considering the impact of nutritional deficiencies in d-la, this may interact with the symptom presentation in ME/CFS and the potential for excess d-lactate accumulation or issues with metabolism. Nutritional deficiencies in ME/CFS may have varied origins, including but not limited to, genetic vulnerabilities, stress, infection, inadequate dietary sources and/or impaired metabolism that are factors involved in the etiology of ME/CFS [153]. Dietary modifications appear helpful for some ME/CFS patients (self-report in [154]) and in clinical case reports [148]. Similarly, dietary treatments and reduced carbohydrate intake were common recommendations for d-la patients (see Additional file 1: Table S1). It would be useful to understand the role of diet as a potential moderating factor (precedent, perpetuating or consequential) in bacterial dysbiosis and d-lactate production in ME/CFS patients and whether this varies for moderately impaired compared to severely impaired (i.e., bedbound) patients.