Interview with Dr. Carlo Cervia-Hasler: Complement system in Long COVID

Interview with Dr. Carlo Cervia-Hasler: Complement system in Long COVID

A Zurich study describing activation of the complement system in Long COVID made headlines in January this year. We spoke to the lead author to understand the impact of the results for those affected.

In a previous post, we briefly summarized the findings from the Zurich study on complement system activation in Long COVID. The study was carried out by Prof. Onur Boyman's Clinical Immunology Research Group. In this interview with the first author, Carlo Cervia-Hasler, we are exploring the results and their implications in more detail.


Carlo Cervia Hasler

Dr. Carlo Cervia-Hasler


First, a brief summary of the study

113 people who had contracted SARS-CoV-2 and 39 non-infected, healthy controls were observed over a period of one year. After 6 months, 40 people showed Long COVID symptoms.

The blood samples of the study participants were tested for >6500 different proteins via proteomics. Proteomics is a technique that can be used to comprehensively research a proteome (the totality of all proteins in a cell, tissue, or organism). It is possible to identify, quantify, and characterize the proteins, thereby gaining insights into the structure, functions, and interactions of the proteins. Through quantification, expression levels of proteins can be determined, and patterns recognized. This allows the activation or inhibition of certain mechanisms to be displayed.

The aim of the study was to find mechanisms and protein expression patterns that are different in people with Long COVID compared to healthy people and people who have had COVID-19 but have completely recovered.

In the study, Long COVID patients showed increased activity of the complement system, which is part of the innate immune system. Healthy and recovered individuals showed no such activation. Certain components of the complement system were different in Long COVID patients (increased levels of soluble C5bC6 complex; lower levels of C7 in the terminal complement complex (TCC)). These changes can result in tissue damage.

Tissue damage, in turn, leads to increased levels of injury markers in the blood. This was confirmed by increased levels of von Willebrand factor and low levels of antithrombin III. In addition, increased platelet aggregation was observed in people affected by Long COVID as well as evidence of antibody-mediated activation of the classical complement system.


In summary, the results indicate that the complement system is particularly active in people affected by Long COVID, that tissue damage occurs and that there is a dysregulation of the antibody response and blood coagulation.

This knowledge can be very helpful for the development of a biomarker test for the diagnosis of Long COVID. It is important to include several of the observed characteristics in a potential measurement.


A research group at Charité questions the results

On March 18, 2024, a research group from Charité in Berlin published a preprint on medRxiv that sharply questions the results of the study from Zurich. These results have not yet been subjected to peer review, i.e., the article has not yet been reviewed by independent researchers.

The Charité researchers criticize the fact that the 6-month Long COVID patients in the Zurich study were significantly older (median age 58) than the people in the control group (median age 35) and had a higher body mass index (BMI). When analyzing a subgroup of study participants whose age and BMI were similar, they could not confirm significant activation of the complement system.

The authors of the Zurich study welcome the investigations of the complement system by the Charité group and emphasize the importance of scientific dialogue and replication studies.

However, they point out that the Berlin group used only one of the proteomics methods (namely mass spectrometry), while the core results of the Zurich study are based on another proteomics method (the SomaScan technology). The different proteomics methods have different strengths, whereas mass spectrometry can measure the total amount of individual complement components, SomaScan technology can measure the amount of complement complexes and total C7. This methodological difference was described in the Zurich study and has now also been confirmed by the Charité.

The authors of the Zurich study also emphasize that differences in age and BMI between the Long COVID patients and the control group were taken into account in their analysis. These factors alone could not explain the observed changes in the complement system.

In the meantime, further studies have also been published that describe complement activation in Long COVID. The results from Charité underline how important it is for a functioning complement system to exactly define what can be measured and find out which measurement method is most suitable. Overall, these publications indicate that an in-depth study of the complement system in Long COVID is worthwhile.


Study Design.png

Number and distribution of study participants.


Questions about the study design:

How were the participants, in particular the healthy control group, selected?

Recruitment of participants began in spring 2020. People with a PCR-confirmed SARS-CoV-2 infection and people who had never had a SARS-CoV-2 infection underwent a clinical check including blood sampling at the beginning of the study, as well as 6 and 12 months afterwards.

Among people who had experienced an acute SARS-CoV-2 infection, there were both mild and severe cases, including people who had been hospitalized due to the infection.

The fact that participants were already selected in 2020 made it possible to ensure that people with a confirmed infection have only been infected once. It was also possible to ensure that the healthy individuals had never been in contact with SARS-CoV-2, which was additionally confirmed by an antibody test.

This data is therefore particularly valuable. A comparable selection would no longer be possible. There are hardly any people left for whom it can be ruled out that they have ever been infected with SARS-CoV-2 and since testing is no longer standard, many people do not know how often they have already had COVID-19.

The study participants were observed from the acute infection or, in the case of the control group, from the negative test result onwards. People in the control group who became infected during the observation period were excluded from analysis. This makes it possible to clearly determine which participants developed Long COVID symptoms as a result of the SARS-CoV-2 infection and which did not.


Did the severity of Long COVID symptoms vary among Long COVID patients in the study? Can we assume that the results are valid for everyone?

The patients included cover a broad spectrum from mild to high severity. However, due to the size of the study (40 patients with Long COVID), it is possible that certain severity levels are not covered. To ensure that all people affected by Long COVID, regardless of their severity, show the same results, more people need to be tested.


Would people affected by Long COVID with an asymptomatic course during the acute infection show the same protein expression pattern?

Although the study participants included people with an asymptomatic course, none of these developed Long COVID. Accordingly, we cannot make any statement about this. However, this would be interesting to look at in a future study. It would also be interesting to see whether different courses of Long COVID disease, i.e., different symptoms, lead to differences in protein expression patterns. Our study has not yet shown any differences.


Is the cohort under consideration large enough to draw meaningful conclusions or is a repetition with a larger number of participants (validation study) necessary?

A validation study is certainly necessary, and we are currently working with collaboration partners to confirm our results in larger groups.


How large would a study have to be or how many participants would be feasible?

There is no "golden number", the important thing is that an additional study is carried out in an independent group of affected people/controls. As far as the desired number is concerned, the more the better, but one has to evaluate what is feasible; a few hundred samples would certainly be a good start.

It is important to consider how well the selected cohort is characterized, i.e., what data is available from study participants and what time point you are looking at, as in how long after the infection are the tests performed.

In the long term, it would also be relevant to do measurements at later time points (after ≥1 year). This would allow us to find out whether the protein expression pattern changes over time, even if symptoms persist. This could provide important information on the period in which a diagnosis is possible using the identified biomarkers.

For a possible diagnostic test, it is necessary to check whether the results change over time. So far, the published study has looked at the time point of 6 months after infection, but samples were also taken after 12 months. These showed comparable results. In addition, we examined people who still had Long COVID-like symptoms one month after the acute infection, but not after 6 months. These did not show the characteristic protein expression pattern observed in the Long COVID patients at 6 months after infection.

In the first study, we went into great detail when analyzing the proteins in the blood samples. Now that we know what we are looking for, we can test more samples in less time in a validation study.


Biomarker Combination for Prediction

In the Zurich study, a combination of these 4 biomarkers was able to reliably determine whether a patient had Long COVID.


The significance of the study results:

How big is the difference in the measured parameters in the 6-month Long COVID group compared to the controls?

The difference is very clear. Especially since we are not just looking at a single parameter, but a combination of several parameters. This allows a clear distinction between people affected by Long COVID and healthy/recovered people on a protein basis. In this study, we built a prediction model based on the results of some of the study participants, which was then tested on the other participants. The agreement was very high. The model was able to clearly predict who had Long COVID and who did not.


Is the difference clear enough to rule out false negatives when using these parameters for a biomarker test?

How many false negatives there are in an actual biomarker test based on our parameters can only be said once the test has been tested and validated in a "real world" scenario.


A large number of diseases have an impact on the complement cascade. Are the identified biomarkers specific for Long COVID and would it be possible to differentiate between Long COVID and other diseases with similar symptoms?

At present, we cannot yet say whether similar protein expression patterns could also occur in other diseases. It is possible, for example, that other post-viral diseases or ME/CFS lead to similar patterns. This could be investigated in more detail in follow-up studies.

Within our study, there were no exclusion criteria regarding comorbidities, i.e., the study participants also included people who suffered from underlying diseases in addition to Long COVID.

At the moment, however, specificity is less relevant for people affected by Long COVID. So far, there is no way to diagnose the disease based on biomarkers in the blood. For now, it is important to find a method that allows a reliable diagnosis. Whether this method is one hundred percent specific for Long COVID can be tested later.


Possible next steps:

What is the next step towards a diagnostic test?

We see our study as a puzzle piece that can help to investigate an underlying mechanism of this disease. On the other hand, it can be an initial indication for the development of a diagnostic test and later a possible therapeutic intervention.

For both, we are dependent on the support of collaboration partners in research and industry. We are currently in the process of finding suitable collaboration partners.

In the meantime, we are trying to better understand the mechanism, add additional measurements at further time points after infection and build new cohorts to prove the results in a larger number of Long COVID patients. We are also constantly working on the technical development of our method to make the measurements easier.


How long would it take to implement such a biomarker test?

If we can quickly find collaboration partners to support us in the development and the development of the diagnostic test proceeds in an ideal way, it will still take at least 2 years before a test is available. How quickly this will then be implemented in everyday clinical practice depends on the hospitals and practices.


How can you find out whether there is a causal relationship between the biomarkers found and the symptoms (i.e., whether the biomarkers are the cause of the symptoms)?

In order to find out whether a causal relationship exists, a large number of model experiments must be carried out. A specific mechanism is first examined in vitro. This means that the behavior of the target proteins (in this case the complement system) is observed under different conditions in artificially produced cells.

If the in vitro experiments confirm the assumptions, further experiments can be carried out in living model organisms, for example mice. However, proving a causal relationship is very complicated and can take several years of research.


In Silico to in Vivo

In research, a distinction is made between in silico, in vitro, ex vivo, and in vivo experiments. In silico experiments are theoretical calculations and considerations that can assess the probability of certain mechanisms. In in vitro experiments, a hypothesis is tested in synthetically grown cell cultures. In ex vivo experiments, the processes to be investigated are observed in tissue taken from a model organism, and finally, in in vivo experiments, the process in question is observed in a living model organism.


There are already drugs that target the complement system. In your opinion, how useful would it be to test such a drug for a systemic treatment that addresses all symptoms of Long COVID in clinical trials?

Drugs that affect the complement system can have dangerous side effects. By downregulating the complement system, parts of the immune system are switched off, thereby significantly increasing susceptibility to certain infections, for example. It must therefore be carefully weighed up whether the potential benefits outweigh the risks.

Apart from that, it would be very interesting to test a known drug with an effect on the complement system in a well-controlled clinical trial.



We would like to thank Carlo Hasler-Cervia for the exciting interview and congratulate him and his colleagues on the study work. We look forward to seeing what happens next!