LASV is a member of the Arenaviridae family and is the etiologic agent of LF, which is an acute and often fatal illness endemic in West Africa. There are an estimated 300,000 - 500,000 cases of LF each year [1–7] with a reported mortality rate of 15%-20% for hospitalized patients. Mortality rates for LF can become as high as 50% during epidemics [3, 8, 9] and 90% in third trimester pregnancies for both the expectant mother and the fetus. Presently, there is no licensed vaccine or immunotherapy available for prevention or treatment of this disease. The severity of LF, its ability to be transmitted via aerosol droplets [10], and the lack of a vaccine or therapeutic drug led to its classification as a National Institutes of Allergy and Infectious Diseases (NIAID) Category A pathogen and biosafety level-4 (BSL-4) agent. The antiviral drug ribavirin has been demonstrated to reduce fatality from 55% to 5%, but only if it is administered within 6 days after the onset of symptoms [1, 8, 9]. There is currently no commercially available LF diagnostic assay, which presents a major challenge for early detection and rapid implementation of existing treatment regimens.

Since 2005, continuous infrastructure improvements at the KGH Lassa Fever Laboratory (LFL) by the National Institutes of Health (United States), the Department of Defense (DoD), the Naval Facilities Engineering Command (NAVFAC), the United States Army Medical Research Institute of Infectious Diseases (USAMRIID), the World Health Organization (WHO), Global Viral Forecasting (GVF) and Tulane University have resulted in the implementation of sophisticated, on-site diagnostic and research capabilities [11, 12]. Currently, LF is diagnosed at the KGH LFL using ELISA and lateral flow immunoassays (LFI) that detect viral Ag. Virus-specific IgM and IgG levels are also determined in serum samples for all suspected cases that present to the KGH LFW. Additionally, the laboratory assesses 14 serum analytes using a Piccolo^® blood chemistry analyzer coupled with comprehensive metabolic panel disks. Flow cytometry powered by a 4-color Accuri^® C6 cytometer performs immunophenotyping, intracellular cytokine and bead-based secreted cytokine analysis on patient sera. These resources contributed to advances in real time diagnosis along with metabolic and immunological characterization of acute LF, thus resulting in a marked improvement in the management of the disease. Herein we present evidence that introduces new insight into humoral and cellular immune responses to LASV that have lead us to reevaluate the role of LASV IgM seropositivity in diagnosing acute LF in suspected cases living in the LASV endemic areas of West Africa. An improved understanding of the natural history of LF will be helpful in guiding future research in diagnosis and treatment.

Historically, LASV viremia has been the primary marker of acute infection as assessed by Ag-capture ELISA, LASV RNA detected by RT-PCR or virus culture (the latter serving as the "gold standard"). Alternatively, elevated LASV-specific IgM has served as a surrogate marker of a recent infection when LASV Ag could no longer be detected [24]. With little known about LASV-specific humoral immune responses and immunopathology, the rationale for considering IgM positivity as acute LASV infection was based on observations in other pathogenic infections in which a drop in antigen levels coincided with increasing IgM titers that class switched to a predominantly IgG status within weeks of initial infection. Given the low rate of IgM and IgG seropositivity in healthy blood donors from the United States, IgM seropositivity in non-Africans recently traveled to LF endemic regions of West Africa is a likely indication of a recent LASV infection. Results presented herein, however, suggest that LASV-specific IgM seropositivity in the absence of viremia should not be considered as a diagnostic marker for acute LF in the regions of West Africa where evidence of a high prevalence of LASV-specific seropositive status exists.

Because Kenema is historically endemic for LASV, we looked to the Northeastern and Southwestern districts of Bombali and Moyamba, respectively, to collect a panel of normal Sierra Leonean serum. These districts lie within reportedly non-endemic areas so exposure to LASV should be low. The donors from these two regions reported normal health at the time of blood draw without febrile episodes in the recent past, and had not travelled to the Eastern provinces where LF is endemic. Contrary to our expectations, a significant percentage in each population had LASV-specific IgM and IgG titers indicating a past infection with LF. Our findings challenged past reports claiming that LF is highly endemic primarily in the Eastern districts of Sierra Leone, and is largely absent in the country's Northern and Southern regions (Additional File 2A, B). It is noteworthy that in the past one year a significant number of suspected LASV cases from Bombali have been referred to the KGH, with several being confirmed as LF [12]. Identification of suspected LF cases has been made possible by increased awareness of the disease by medical staff in Bombali district hospitals and the availability of rapid diagnostics provided by the VHF consortium.

These surprising results prompted us to reevaluate the categorization and analysis of Ag and IgM status in a suspected LF patient database compiled at the KGH LFL between 2006 and 2011. To determine whether IgM status could be used as a diagnostic marker, we compared the odds ratios of a group of LASV Ag+ patients and LASV Ag- patients. This revealed that, regardless of IgM status, the fatality rate was significantly higher in Ag+ patients compared to Ag- patients (Ag+IgM+ v. Ag-IgM+ OR = 4.33, p < 0.01; Ag+IgM- v. Ag-IgM- OR = 5.36, p < 0.01). Additionally, we found that positive LASV IgM seropositivity is not a reliable marker of acute LF base on the adjusted odds ratio of 1.13 in Ag-IgM+ (N = 171) versus Ag-IgM- (N = 244) in suspected LF patients (p = 0.61, Table 1). Conversely, Ag+ status, irrespective of IgM status, was indicative of a poor outcome, with approximately 55% of patients succumbing to LF. This percentage is higher than the historical and consistently reported death rate of approximately 15-20% because we eliminated a large population of patients that could not be considered as acute infections with LASV [25–28]. We believe that 55% is a more accurate fatality rate based on the prolonged LASV-specific IgM response found in convalescent patients months to years following LASV infection (Figure 2) [11, 12]. Our data show that detectable LASV viremia assessed by Ag seropositivity is highly correlated with acute viral infection, while IgM shows no such correlation.

There are several potential interpretations for sustained LASV-specific IgM titers [29–33]: 1. IgM may be indicative of a recent infection for which class switching has not fully occurred; 2. A prolonged and as of yet largely uncharacterized inhibition of class switching could be prevalent in LASV infections (Grove, Garry et al., unpublished data); 3. Sustained IgM titers could be generated by IgM+ memory B cells; 4. Impaired CD4+ T helper lymphocyte function during LASV infection, with long-term impact on class switching may be central to humoral and cellular immunity aspects in LF; (5) Individuals may be infected with LASV but clear viremia without treatment, and seek medical intervention too late for detection of virus antigen by LFI, ELISA, and nucleic acid by PCR.

Evidence for these mechanisms can be found in the literature. Non-neutralizing virus-specific IgM were recently analyzed and deemed crucial for impedance of viral persistence in LCMV infection (the prototypic Old World arenavirus), suggesting that IgM-producing cells have remained largely obscure and underappreciated in the characterization of arenaviral infections [34]. Additionally, the late appearance of neutralizing antibodies in arenaviral infections has been tied to high viral antigen-to-B cell ratios and low T cell help, which resulted in a normal IgM response but reduced the efficiency of class switching [35]. Lowering the antigen-to-B cell ratio and increasing T cell help resulted in rescuing of class switching and emergence of neutralizing IgG specificities. Thus it is possible that a normal early IgM response in LASV infections is followed by an impaired T helper cell response with sustained IgM production and eventual maturity of the producing B cell subset into an IgM+ B cell memory population. The role of IgM and T-helper lymphocytes have been studied in controlled cynomolgus macaque models of SIV infection [36]. These studies noted a strong inverse correlation of the immunoglobulin and CD4+ T helper cell counts after the primary peak of IgM response, reflecting the prevalence of mature plasma cells that have not undergone class switching. Moreover, a strong correlation was observed in the same studies between pre-infection immune status and disease progression. It is therefore possible in arenaviral infections, as in the SIV model, that a normal, pre-infection CD4+ T helper cell threshold may regulate normal B cell responses, with concomitant class switching, emergence of strong neutralizing IgG titers, and quiescence or elimination of short lived IgM-producing plasma cells. We have recently reported on two cases of severe LF [11, 12] that registered increasing IgM titers throughout hospitalization and for months into convalescence. Virus-specific IgMs were invariably paralleled by rising, albeit temporally delayed, IgG titers. We have observed similar circulating Ig patterns in numerous other LF cases, with rising IgG titers following an initial IgM response. Interestingly, in the 21 NL FI IgM+ subjects we did not record LASV-specific IgG titers, despite prolonged hospitalization and multiple sera testing for a subset of patients (Figure 2 A, B). Therefore, it appears that in this group of patients LASV-specific IgM responses were indicative of a previous infection with muted class switching.

In the Ag positive subset of patients, remarkable differences were noted in cytokine and metabolite levels that could be used as early prognosticators in LF. Our results have confirmed historical reports demonstrating significant hepatic and renal dysfunction in LF, namely the high levels of ALP, ALT, and AST [30–39]. However, not all of our cytokine and metabolite observations follow the historical observations from the literature. Mahanty et al. previously reported that elevated levels of IL-8 and IP-10 correlated with positive outcomes in LF [24], while we observed that low levels of IL-8 and other cytokines to be correlates of survival outcome. The frequent exposure of populations to parasitic infections is a likely factor in the elevated levels of recorded IL-8 in the sampled normal and non-Lassa febrile groups, a phenomenon not commonly observed in U.S. blood donors (Figure 4). Although there was not a significant reduction in the overall levels of IL-8 between normal donors and fatal LF patients, a significant drop in the levels of this cytokine was observed in all patients who survived LF (excepting a single, high outlier). Thus our data from this report and others [37–39] suggest that a regulated quiescence of the inflammatory response, in addition to a robust humoral immune response may be central to a successful outcome in symptomatic acute LF. In addition to IL-8, low levels of IL-6, IL-10, MIP-1β, and CD40L were predictive of survival. A subset of individuals presenting with a non-Lassa febrile illness that may have registered LASV-specific IgM titers were not included in our acute LF study group, but were included in the acute LF population studied by Mahanty et al. [24]. The differences between our studies and those by Mahanty et al. [24] and others may therefore rest on the parameters employed in classifying acute LF.

Our analysis of a nonfatal non-Lassa febrile illness (NF NL FI) study group revealed levels of IL-8 similar to those in the LF NF and LF F groups, despite observing increased levels of the inflammatory mediator in some subjects (Figure 4). The only inflammatory marker analyzed in our studies that differentiated LF from other FIs was IL-6, which was increased in LF F relative to all other comparison groups (Figure 4). Interestingly, in the NL FI IgM+ (and IgG- Ag-) group we observed reduced levels of IL-8 and elevated levels of MIP-1β, as well as metabolic indicators that did not significantly differ from those in LF NF. A comparable profile of inflammatory mediators was observed in a study of LF pathogenesis in cynomolgus macaques, which revealed that elevated levels of IL-6 conferred a poor prognosis, while IL-8 and IL-10 responses were largely absent [40].

Together, these data indicate that upon admission to the hospital with suspicion of LF, a profile of low to moderate IL-6, -8, -10, MIP-1β, CD40L, BUN, ALP, ALT, and AST levels predict a positive outcome following treatment with a full regimen of ribavirin, fluids management, antibiotics, and other appropriate medical intervention. Conversely, AST levels greater than 2000 U/mL are almost always indicative of a poor survival outcome. In addition to high AST, combined elevated IL-6, -8, -10, BUN, ALP, ALT, and reduced tCO₂, Ca²⁺, RANTES, and CRP levels provide a statistical basis for poor prognosis. Two recent extensively-characterized reports of hemorrhagic LF in Sierra Leone with positive outcome presented to the KGH LFW with AST > 2000 U/mL and low levels of IL-6, -8, and in one case elevated IL-10 [11, 12]. Based on the data presented herein, the prognosis for both patients would have been poor, yet, with a relatively short timeframe from onset of symptoms to presentation to the KGH LFW (6 and 7 days, respectively) and proper medical interventions, both patients survived. The data collected in these studies generated statistically relevant correlates of LF outcome, but have also exposed gaps in our current understanding of relevant biomarkers for LF. The observed pronounced quiescence of the inflammatory response in surviving LF patients may present new opportunities for future disease treatment and management. It is conceivable that administration of anti-inflammatory drugs to quiesce the cytokine storm observed in LF may complement the beneficial effects of reducing viral loads with ribavirin treatment, thus increasing survival rates in acutely infected patients.

Limitations of this study include lack of randomization, limited number of fatal Lassa cases, potential patient by treatment interaction, and the limited duration of the study. Random assignment of subjects to comparison groups was not possible due to the uncontrollable nature of the outcome, and random selection of patients was impractical due to the limited number of fatal Lassa cases. As a result, some of the results that we observed may be due to inherent characteristics of our study subjects. This is perhaps most evident in the 13 normal donors from Bombali that registered greater than 90% IgG reactivity to LASV antigens. It is important to note that these limited studies were not designed to ascertain levels of IgM and IgG seroprevalence in Sierra Leonean populations. Instead, through the continuous analysis of incoming sera from patients and random collection of normal samples in two districts outside historically endemic regions, it was established that LASV infections in humans might be more prevalent across Sierra Leone than previously reported. In the case of Bombali donors we may have collected sera from a cluster of individuals who may have had an unknown exposure bias to LASV, thus registering high levels of IgG (and possibly IgM) to viral antigens. We suspect our study subjects are more likely to have multiple exposures to LASV than subjects from previous studies due to the increasing prevalence of the disease. Due to the small sample sizes for each of the comparison groups, we were unable to adjust for confounding variables, which makes the generalization of these findings inappropriate, particularly for those outside of Sierra Leone. Although all of the patients in this study were subjected to the same treatment protocol, it is unknown whether the treatment was administered in a consistent manner and whether patients responded differently to the ribavirin treatment. A randomized trial on the effect of ribavirin would be beneficial to future studies on Lassa fever. It is worth noting that there is a temporal component for the outcomes investigated in this study that is not well understood.

Due to our characterization of a prolonged IgM response in convalescent LF patients, high prevalence of IgM seropositive in healthy normal controls, and the failure of most IgM only suspected cases to display a dysregulated metabolic and inflammatory cytokine profile similar to LASV-specific Ag+ patients regardless of IgM status, we suggest that the traditional paradigm for diagnosis of acute LF in West Africa should be reconsidered and changed. Rather than diagnosing an Ag+ and/or IgM+ result as acute LF, only in the subset of patients displaying LASV viremia, as determined by LFI and Ag-capture ELISA, with confirmation by RT-PCR should be definitively categorize a patient as being acutely infected with LASV. Patients that presented with symptoms indicative of potential LF and who do not test positive for viremia by any of the three methods employed while presenting with IgM and/or IgG titers should not be categorized as acute LF. Unfortunately, ultimate diagnosis of the non LF febrile illness for these patients will often not be possible due to low resources and lack of diagnostics for additional suspected infectious agents. Though, the new prognostic immune correlates of LF identified in this study will allow the scientific community to better understand, monitor, and possibly treat and prevent the disease.