Why Does COVID-19 Cause Severe Illness in Some Patients but Not Others?

Why some individuals with COVID-19 experience symptoms no worse than the flu while others suffer severe, sometimes deadly outcomes has puzzled researchers throughout the pandemic. Even among hospitalized patients, there is a wide divergence of clinical outcomes. A new study harnesses big data and bioinformatics to explore how the immune system can fight off the virus in some cases and fails in others.

In May 2020, researchers at 15 biomedical centers throughout the U.S., including Yale, launched an observational, NIAID-funded study called IMPACC (IMmunoPhenotyping Assessment in a COVID-19 Cohort) to profile COVID-19. IMPACC is using a systems immunology approach developed by the Human Immunology Project Consortium (HIPC) of the National Institute of Allergy and Infectious Diseases (NIAID), part of NIH as well as by other NIAID-funded research networks. The HIPC is directed at Yale by IMPACC members Ruth R. Montgomery, PhD, professor of medicine and of epidemiology (microbial diseases), and David A. Hafler, MD, chair and William S. and Lois Stiles Edgerly Professor of Neurology.

Now, a team co-led by Steven H. Kleinstein, PhD, Anthony N. Brady Professor of Pathology, has analyzed the first set of deep immunophenotyping data collected from the IMPACC cohort of over 1,100 hospitalized individuals infected with SARS-CoV-2. The research is designed to help scientists predict early on who is most likely to suffer severe infection and potentially death, while also identifying aspects of the immune response that might be modulated to help at-risk patients experience better outcomes. The team published their results based on the first half of the cohort (>15,000 samples from 540 participants) in Cell Reports Medicine on May 23. Follow-up studies on the rest of the IMPACC cohort are expected to follow soon.

“We wanted to identify differences in the immune responses in participants with milder disease who were generally discharged from the hospital quickly versus those with a more severe disease course to understand what might be driving those different trajectories,” says Kleinstein, who was the study’s senior author.

Previous smaller studies had identified potential links among various aspects of immune responses and disease outcomes, but they didn’t look at the broader picture. “There was not a lot of information on how these different aspects of the immune response correlated with each other because each study that would measure one aspect didn’t measure others,” says Kleinstein. “We wanted to look at how they all tied together.” Furthermore, most of these studies were cross-sectional and didn’t track how immune responses unfolded over time within individual participants.

The team’s latest publication is an observational study of hospitalized adults with a PCR-confirmed SARS-CoV-2 infection. In total, they enrolled 1,164 participants into the IMmunoPhenotying Assessment in a COVID-19 Cohort (IMPACC). The researchers tracked the participants over the course of their hospitalizations, and followed up multiple times up to a year post-discharge requiring tremendous effort from many colleagues (led at Yale by Albert C. Shaw, MD, PhD, professor of medicine (infectious disease)). During these visits, the researchers carried out 14 distinct immune profiling assays on samples from each participant with most assays conducted in one core lab to maximize consistency of data generation.

These assays quantified multiple immune parameters using several core approaches. First, they looked at serology. Through analyzing the blood, the researchers could study the antibodies produced in response to the virus as well as auto-antibodies that may act against the body’s own proteins. They also looked at the proteomics, including the proteins circulating in the serum and plasma that reflect the immune status of the participants. They studied circulating metabolites and cell-type frequencies. Furthermore, they analyzed the genomics, including all gene expression and germline genetic variations in each participant. In addition to blood, many of these assays were also used to profile the nasal epithelium, the port of entry for the virus.

In an earlier study, the team looked at the participants’ respiratory status over the course of hospitalization and divided them into five clinical trajectory groups, ranging from milder hospitalization cases to mortality. In their latest paper, participant immune responses were analyzed for associations with the clinical trajectories. This offered two major types of insights. First, within the first 72 hours post-hospital admission, the team found distinct immunological states associated with the illnesses’ future trajectories. Factors linked to disease outcomes included viral load, frequencies of certain cell types such as cytotoxic natural killer (NK) cells, various inflammatory states, and markers of myocardial damage.

Second, the study offered a look into how these immunological states evolved over time. For example, in participants who died of the infection, viral loads persisted and even started to rise over time. By contrast, viral loads steadily declined in those with milder infections. The team also saw the activation of cell types such as cytotoxic NK cells in the milder cases as participants’ immune systems ramped up in response to the infection, while these cells would decrease in the more severe groups. Furthermore, participants who were sickest had higher levels of inflammatory markers upon hospitalization that increased over time, while patients in the milder groups showed an opposite pattern.

As tens of thousands of people around the world continue to catch COVID-19 each week, studying the diversity of immune responses will help researchers learn how to better manage new infections. It will also help improve the understanding of host immune responses to viruses more broadly. “We’ve developed a very rich data set with over 130 million data points,” says Kleinstein. The team hopes eventually to use this work to develop predictive models of clinical outcomes. Kleinstein is especially interested in using the data set to better understand the two groups with the worst trajectories of disease, which included participants who suffered severe disease and recovered and those who died of the infection. In future studies, he and his team hope to uncover predictive markers that distinguish those who survived from those who didn’t.

“The scale and richness of this dataset enable us to investigate this extremely challenging task. Furthermore, the data offer the opportunity for understanding the temporal coordination between different immune signatures, shedding light into the mechanistic insights distinguishing symptom heterogeneity and differences between the two groups of patients [Kleinstein] mentioned,” says Leying Guan, PhD, assistant professor of biostatistics, who helped in designing the analysis protocol in this published work and is leading a follow-up project investigating early risk predictors and temporal coordinations of different immune components using the full cohort.

The new study emphasizes the first 28 days post-infection, says Kleinstein. But for many of those who have suffered from COVID-19, the symptoms linger for weeks, months, or even years. Next, the team hopes to use the IMPACC cohort to understand associations between immune response and post-infection symptoms. “Overall, we’re hopeful to not only gain an understanding of how to better treat the acute infection, but also of how the course of disease and immune state are related to persistent symptoms,” he says.

Contributing to the Yale IMPACC study are:

Anderson Brito

Ernie Chen

Chris Cotsapas

Denise Esserman

Nathan Grubaugh

Leying Guan

Jeremy P. Gygi

David A. Hafler

Steven H. Kleinstein

Anna Konstorum

Ruth R. Montgomery

Shrikant Pawar

Albert C. Shaw

This research was funded by NIAID through grant numbers 5R01AI135803-03, 5U19AI118608-04, 5U19AI128910-04, 4U19AI090023-11, 754 4U19AI118610-06, R01AI145835-01A1S1, 5U19AI062629-17, 5U19AI057229-17, 755 5U19AI125357-05, 5U19AI128913-03, 3U19AI077439-13, 5U54AI142766-03, 5R01AI104870- 756 07, 3U19AI089992-09, 3U19AI128913-03, 3U19AI1289130, U19AI128913-04S1, and R01 AI122220.