Chronic Fatigue Syndrome (CFS) is a complex clinical condition of unknown etiology, characterized by persistent or intermittent fatigue that is not the result of recent exertion and does not improve with rest, resulting in a significant reduction in the patient's previous normal activity[1]. Classical diagnostic criteria for CFS overlap with Myalgic Encephalomyelitis (ME) and require these symptoms to be present for at least six months and concomitant to at least four accompanying symptoms, among them: impaired memory, adenopathy, myalgia or polyartralgia[2]. The World Health Organization acknowledged ME/CFS as a disease of the nervous system (ICD G93.3)[3]. However, CFS is a multi-system disease, in which neurological disorders are accompanied by altered immune, musculoskeletal, endocrine and cardiovascular systems[4, 5]. Research efforts have recently provided a new International Consensus Criteria for ME[3]. However, the search for the etiology(ies) and the pathogenic mechanisms of CFS has unsuccessfully reviewed several viral hypotheses including herpesviruses or retroviruses as potential triggers of the disease[6–8]. The potential role of pathogens in the CFS field has boosted research in the immunological sides of the illness[9–11]. Indeed, the link between the immune system and CFS has been explored since the early 90’s[12], and is supported by the coincidence of the onset of symptoms with viral infections[13], the persistence of several infections in CFS individuals[10, 14–17], the beneficial effect of treatment of human herpesvirus 6 and Epstein-Bar virus infection in CFS symptoms[18] and the reported role of autoimmunity[19, 20] Moreover, other features of ME/CFS, such as mild inflammation, oxidative/nitrosative stress, mitochondrial dysfunction and the presence of autoimmunne responses[4] may alter immune function and phenotype.

However, the characterization of the immune status of CFS individuals has frequently yield contradictory results. Early work associated CFS with a general status of immune activation assessed by CD38 or HLA-DR expression in CD8 T cells[12]. However, other authors found similar expression of these markers in CFS and healthy individuals[21], while lower expression of the activation markers CD69 or soluble CD26 has also been described as a feature of CFS[22, 23]. Similarly, B-cell function[24], B-cell mediated autoimmunity or unbalanced cytokine network have been linked to CFS[9, 25–27] again with inconclusive results. Altered numbers of NK subsets, defined by CD56 or CD16 expression, and an impaired NK-cell lytic activity have been more consistently associated with CFS[24, 28–32], although controversial data have been also reported[33] and a consensus in relevant NK cell markers is still lacking.

At least some of these immune features described in CFS may be related to active, poorly controlled viral infections, which differently modulate immune responses and may produce immune hyperactivation/exhaustion, as widely reported for HIV[34]; may contribute to immunosenescence, as postulated for CMV[35]; or may cause a status of immune anergy, as described for measles virus[36]. In this context, we sought to compare the phenotype and function of different immune cells between healthy and ME/CFS individuals. The study was designed in 2010 and could not include new definitions that currently allow for the separation of ME and CFS affected individuals[3]. However, our data point to an unaffected B-cell compartment, a biased NK-cell phenotype and a poorly responsive T-cell compartment as the main immune features of ME/CFS affected individuals.

The immunological status of individuals suffering from CFS has yielded heterogeneous results[12, 17, 31, 51, 52]. Clear examples are data from different CFS cohorts analyzed in a single laboratory[32]. These observations may be the consequence of an intrinsic heterogeneity in the classification of CFS affected individuals[3], or could be related to the presence of intercurrent infections in the individuals recruited in the different studies. In particular, the prevalence of infections by herpesviruses or enteroviruses[15], which have been described more frequently in CFS individuals[18, 53], are known to modulate immune phenotype[46]. We have performed a wide screening of the phenotype and function of B, NK and T cells in CFS. In contrast to other studies, our main inclusion criterion was focused on the lack of active infections, rather than on the CFS-related comorbidities. Although this could be a limitation, as comorbidities may also affect immune status, we believed that these criteria might provide a more homogeneous immune profile of CFS.

Our data suggest that most differences between CFS and healthy controls were observed in NK and T cells; while the B-cell compartment showed similar composition in both groups. Importantly, these differences could not be associated to polypharmacy or comorbidities (Additional file1: Figure S3 and data not shown), although the sample size of our study limited the statistical power of these analyses. The lack of B cell alterations observed in our study contrasts with the active role of B cells and autoimmune responses in CFS that have been highlighted by the recent clinical use of Rituximab[19]. The possibility that B-cell alterations are restricted to tissue B cells may explain this apparent contradiction. However, it should be noted that in contrast to NK and T cell markers described herein, parameters of B cell phenotype showed heterogeneous values and seem to be more affected by antioxidant/analgesic treatments. This could be a second limitation of our study; thus, a more detailed analysis of B cell phenotype and function using larger cohorts will be required to fully understand the role of these cells in CFS.

A different scenario is observed in NK cells. In this case, several markers showed consistent alterations in CFS individuals and illustrate a skewed NK cell population with high CD69 and low CD25 expression, a paradoxical phenotype that has been described in acute influenza infection or vaccination[54] and that is in clear conflict with recent data reporting low CD69 expression in NK and T cells from CFS affected individuals[22, 23]. Of note, this latter work measured CD69 expression done after in vitro stimulation while our data were obtained using freshly obtained unstimulated cells. Thus, although apparently in conflict, both data may reflect different aspects of a deregulated CD69 expression in CFS. We have also observed increased levels of NKp46 expression in CFS individuals. Interestingly the expression of this receptor has been recently linked to T-cell responses in murine models[55]; however, no clear association could be found in our cohort. The phenotype of NK cells is controlled by genetic, epigenetic and environmental factors and probably summarizes the infectious history of an individual in a poorly-understood form of immunological memory[56]. Specific phenotypes have been associated to infectious agents, such as CMV or influenza[54, 57]. Although it could be tempting to speculate on a common infective history in our cohort of CFS individuals, the analysis of their serological status for some viruses (CMV, EBV, parvoviruses) did not show a clear link between past infections and current NK-cell phenotype. Probably, wider studies are required to answer this relevant question and to confirm the association of NK-cell phenotype with impaired lytic activity, which has been described in larger cohorts but failed to reach significant differences in our limited analysis, probably due to the small subset of samples analyzed or to the use of EDTA as anticoagulant. NK function decreases with age, a phenomenon associated with NK immunosenescence that can be evidenced by the increased expression of CD57, present in terminally differentiated NK cells[58]. However, there is no consensus on the role of CD57 in immunosenescence of NK cells and CFS has been associated with low expression of CD57[59]; an observation partly confirmed in our cohort.

The most unanticipated data in our cohort of CFS individuals is related to T-cell phenotype and function, which could be defined as a general hyporesponsiveness. These data contrast with descriptions of high T-cell activation in CFS[12], but are consistent with other reports describing reduced CD8 cytotoxic activity[9]. Again, the presence of active viral infections at the sampling time may be a source of heterogeneity. Alternatively, the leakage of bacterial products from gut may also determine T-cell activation[60, 61]. To explore this possibility, we analyzed plasma levels of sCD14 In healthy and CFS individuals, showing similar median levels (4.5 and 4.7 μg/ml, respectively, p = 0.44, data not shown), suggesting that gut leakage is not a major contributor to immune alterations in our cohort.

Our data may suggest a general default of T-cell function that can be observed using different makers in both CD4 and CD8 subsets. While CD4 T cells show lower Ki67 staining and ex vivo proliferation, CD8 T cells showed no proliferative differences, but lower levels of CD56 expression, effector (CCR7^–CD45RA^–) cells, CD38⁺ cells and higher expression of CD5, a marker of anergy associated to continuous antigen exposure[50]. Furthermore, this general status of T-cell hyporesponsiveness seems to be unrelated to immunosenescence, since no differences in CD57 or no increased levels of CD27^- or CD28^- cells were observed between groups. A key factor in the control of T cell responses is the function of Treg cells[62], which are significantly increased in our cohort of CFS individuals confirming data from another recent study[9]. Although no clear correlation was observed between Treg frequency and other markers of NK cell or T-cell phenotype; an active role of Treg could be supported by reported data on key mediators of Treg action, such as TGF-β[63], that seem to be also upregulated in CFS individuals[64, 65].

The potential use of the immunological markers identified in this study was explored in cluster analyses. While the best resolution required both T and NK cell markers, we also identified a robust combination of NK cell markers that may be useful for diagnosis. This combination includes CD25, CD69 and NKp46 expression in CD56 + CD16+ cells and CD56 expression in CD3⁺ cells, suggesting that a five-color flow cytometry strategy including these markers may be useful for diagnostic purposes. However, a wider range of parameters including FOXP3 and Ki67 expression in CD4 T cells improved specificity. The use of these potential combinations as diagnostic tools requires validation in further studies, including both larger cohorts and a wider range of CFS clinical status.

In conclusion, CFS individuals analyzed in this study show no differences in B-cell compartment, skewed NK cells and poorly responsive T cells. The observed immunological defaults do not provide any causative link to the illness, but could explain some of the symptoms and in particular the poor control of viral infections reported in these individuals[15, 53]. However, some of these markers along with other previously described, such as NK function or DPPIV[22, 66] may be useful for the characterization of CFS. However, major roadblocks still exist to reach a reliable combination of immune biomarkers for CFS. First, the potential different etiologies or comorbidities of CFS and second the clear identification of the target population (ME, CFS or both). In addition, immunological markers may reveal only a part or the complex pathogenic spectrum of ME/CFS. Most likely, immune features in combination with a detailed analysis of intercurrent infections and previously described[9, 22, 25, 29, 66, 67]neurological and metabolic disorders may provide the clues to define a full set of markers helpful for our knowledge of CFS pathogenesis and for its clinical management.