To the best of our knowledge, this is the first study in healthy volunteers to investigate the cytokine, cellular, and clinical response to repeated intravenous LPS challenges over a year apart. We report three main findings: First, we found that the biochemical (i.e., cytokines and CRP) and cellular responses showed statistically significant intra-individual correlation, which supports the reproducibility after repeated administration of LPS over an extended period of time. Second, and in contrast to the previous finding, clinical symptoms did not show consistent reproducibility. Third, the average magnitude of cytokine response, especially IL-6 and TNF-alpha, tended to be weaker after the 2nd LPS challenge, potentially implying a long-lasting endotoxin tolerance. The magnitude of the cellular response and clinical response was similar between the 1st and 2nd LPS challenge.

Several studies showed that repeated pulmonary LPS challenges produce consistent intra-subject cytokine and cellular responses after intervals of 4 to 5 weeks [14,15,16]. Similarly, in intravenous LPS challenges separated by at least 5 weeks, the extent of cytokine release within each subject was reproducible [6]. The present study indicates that IL-6, IL-8, TNF-alpha, and CRP show a strong intra-individual correlation after more than one year, a time span that has not been investigated so far.

Various previous studies explored the dynamic cellular changes in healthy volunteers following intravenous infusion of LPS, mostly focusing on one subset of leukocytes [17,18,19]. To our knowledge, this is the first study using a holistic immune profiling approach to describe the cellular dynamics after LPS infusion in healthy volunteers. Consistent with the available literature, our data showed that endotoxemia has a significant impact on virtually all circulating leukocyte populations, mainly characterized by an increase in neutrophils and a decrease in lymphocytes, monocytes, and other subsets of the innate immune system. Moreover, we observed that the extent of leukocyte subset changes after 4 h was stable within the individuals after the repeated LPS challenge, except for monocytes. We hypothesize that the recruitment and migration of leukocytes are significantly determined by individual constant factors, most likely (epi-)genetic factors or other constant factors such as sex or slowly changing factors such as age [20]. As previously described by Wegner et al. [21], females exhibited numerically greater AUCs and peak concentrations for all four cytokines during the 1st and 2nd LPS challenge, albeit without reaching statistical significance in the small sample size of this study (5 males vs. 4 females). Despite the lack of statistically significant sex differences, numerical differences may still have influenced the analyses presented in this work.

Monocyte dynamics did not show significant reproducibility, possibly because monocytes are highly plastic cells characterized by marked heterogeneity and versatility [22]. As a result, monocytes may be more influenced by environmental factors, compared with neutrophils or lymphocytes, which exhibited more stable long-term behavior and less pronounced turnover [22]. Since monocytes play an important role in the response to LPS, it could be hypothesized that the variability of the monocyte response could contribute to the variability of the clinical response.

The exact mechanism of cellular dynamics during endotoxemia is unknown, but the pronounced changes in virtually of subsets are likely caused by re-distribution. It has been shown that LPS administration increased endothelial activation markers including vascular cell adhesion molecule1 (VCAM1), thrombomodulin, and P-Selectin, which may result in the extravasation of particular subsets [23]. Cells may also be removed from the bloodstream by adhesion to and rolling on the endothelium and thus enter the so-called margination pool [24, 25], which is associated with systemic inflammation [26]. On the other hand, systemic inflammation induces the release of neutrophils from the bone marrow, which may help to explain the neutrophilia after LPS exposure [27, 28].

Our study did not show a consistent reproducibility of the clinical response to intravenous LPS. The clinical manifestations are ultimately what inflammation research is interested in, as biochemical and cellular processes are merely biomarkers. Notably, clinical responses are difficult to quantify, partly subjective, and altogether more complex as they are the joint result of a multitude of determining factors. It is therefore not surprising that it is more challenging to show intra-individual reproducibility of the clinical response to LPS than that of biomarkers like cytokines. Moreover, the lack of reproducibility may indicate that clinical symptoms have – in addition to cytokine and cellular factors – even more sources of variability, such as environmental factors, day-to-day variability, or subjectiveness. In addition, 1 of 9 subjects received paracetamol in the 1st round, and 5 of 9 subjects received paracetamol in the 2nd round, despite a somewhat lower IL-6 response. Paracetamol was typically requested when symptom severity peaked. However, it is not known whether this prevented further worsening of symptoms and vital signs and thus introduced relevant confounding. We only observed a statistically significant correlation between the self-reported severity of nausea and myalgia. Whether these correlations are a product of chance (i.e., a type I error) or a slight indicator of intra-individual reproducibility of clinical responsiveness remains unclear and cannot be deduced from our data.

Numerous studies have demonstrated that repeated LPS exposure results in less pronounced cytokine responsiveness in healthy volunteers. However, the durability of this effect, which is commonly referred to as endotoxin tolerance, is undetermined [5, 29].

In a previous study at our site, we found lower levels of circulating cytokines following a repeated intravenous LPS challenge after 6 weeks [8]. Likewise, Janum et al. also found a lower cytokine release following a second intravenous LPS challenge in a crossover trial with a washout period of 4 weeks [30].

In the present study, we found a significantly reduced release of IL-6 and TNF-alpha after the 2nd LPS challenge, possibly indicating long-lasting endotoxin tolerance. IL-10, IL-8, and CRP were also numerically lower after the 2nd LPS challenge, but without becoming statistically significant. The kinetic profile, which was characterized by peak concentrations at around 2 h followed by a clearance until 6 to 10 h after LPS administration, remained similar during the repeated endotoxemia. Regarding the cellular response, we found few statistically significant signals indicating a lower or higher responsiveness at the 2nd LPS challenge. Given the large number of comparisons of relative and absolute cellular subsets and the lack of a consistent and plausible pattern in these differences, we do not believe that our data robustly supports an attenuated cellular response after repeated LPS infusion.

Although these results should be interpreted with caution, we cannot rule out some degree of endotoxin tolerance, which may be associated with an attenuated cytokine response, even after an extended period of time. As Netea et al. pointed out, current evidence suggests that the reprogramming of the innate immune system can last for periods of time on the order of years [3, 29, 31]. While the exact mechanisms of immune tolerance are still being explored, the important role of epigenetic reprogramming, which could also help explain the long-lasting effect that extends well beyond the lifespan of circulating mature leukocytes, is increasingly recognized [3, 32]. Subsequently, responsiveness to LPS may also be influenced by previous bacterial infections.

LPS is considered as a potential adjuvant for vaccines, but its unmodified form is limited by its toxicity. LPS has therefore been detoxified to monophosphorylated lipid A (MPL) [33] to stimulate TLR4-mediated responses while reducing systemic inflammatory toxicity. For example, MPL is currently used in the licensed shingles vaccine Shingrix®. On the other hand, meningococcal vaccines use outer membrane vesicles containing lipooligosaccharides – the Neisseria meningitidis analogue of LPS that lacks the repeated residues of the O antigen [34] – which are also chemically detoxified [35]. In contrast to the intravenous LPS challenge, the LPS components in these vaccines are designed to stimulate the adaptive immune response without inducing pronounced systemic inflammation. Given the observed tolerance to repeated intravenous administration of LPS, it is conceivable that this tolerance could also be found to modified LPS or lipid A, potentially impairing immunogenicity after vaccination. Considering the modest degree of LPS tolerance observed in our study, it is unlikely to play a clinically relevant role by reducing immune responses or even protection after extended time periods. Since TLR4 signaling and subsequent cytokine responses depend on the modifications of LPS [36], strain-specific differences in the spectrum of endotoxicity should also be considered when interpretating these data [37].

First, this was an exploratory study in which no adjustment for multiple comparisons was made. Because of the large number of statistical tests, our results are at considerable risk of false positivity and should be interpreted as hypothesis-generating only. Second, quantification of cellular subsets using CyTOF was performed at 0 and 4 h, which only provides an incomplete temporal resolution of the complex cellular dynamics following endotoxemia. Third, the number of subjects was small (9 for the cytokine and clinical response and 8 for the cellular response). However, establishing statistically significant correlations with such a small sample size is only possible in the case of strong and thus potentially meaningful correlations.

Fourth, seasonality is another important source of variability [20], for which controlling was not feasible for us. Notably, technical and study design-related sources of variability would bias toward the null hypothesis and against the finding of a statistically significant correlation. Finally, we acknowledge that we studied statistical differences within a model for systemic inflammation in young healthy volunteers. While our findings may help to hypothesize immunological mechanisms, the clinical relevance of the observed results is unknown without confirmatory studies.