Diabetes in kids after covid infection, and an explanation for long covid

 Two interesting articles came out on long Covid, one finding increased type II diabetes in kids and the other finding residual spike protein in many patients that might lead to continued immunologic response and long covid

 

A retrospective cohort study found that new diagnoses of type II diabetes were significantly increased up to six months after Covid diagnosis (see covid diabetes in kids JAMA2024 in dropbox, or doi:10.1001/jamanetworkopen.2024.39444)

 

Details:

--a retrospective cohort study using electronic health records from the TriNetX Global clinical analytics platform used specifically for research (a de-identified database of over 100 million patients from 60 large healthcare organizations across the country and representing diverse geographic areas, self-reported race and ethnicity, age, income, and insurance types including Medicare and Medicaid), between 2020-2023

    -- 613,602 patients were evaluated who were aged 10 to 19 without pre-existing diabetes: 306,801 with Covid and 306,801 with other respiratory infections (ORIs) but no documented Covid

-- mean age 15 years old, 53% female, 14% Hispanic/Latinx, 57% white/19% Black

-- BMI was documented in 18% having Covid and 23% having other respiratory infections, 8% of both were above the 95th percentile for age, 5% were between the 85th and 95th percentile

-- primary hypertension 1%, family history diabetes 1%, hyperlipidemia 1%, all other measured diseases (an array of autoimmune diseases as well as polycystic ovary, low birth weight newborns, and long-term systemic steroids) were all less than .05%

    -- there was propensity score matching to equalize the variables above that were significantly discordant between the two groups

 

--primary outcome: comparing the incidence of type II diabetes, comparing risk ratios at one, three, and six months after the index covid infection versus other respiratory infection (ORI)

 

 Results:

 

 

-- so, there are a few general results here:

    -- comparing the risk of new diagnoses of type II diabetes: a pretty consistent 50-60% increased risk in those having Covid versus those with ORI diagnoses

    -- the increased risk in those people who were classified as overweight or obesity: pretty consistently a bit more than twice as common in those with Covid versus ORI diagnoses

    -- the increased risk in those who were hospitalized: 2 ½ to 3 times as high in those with Covid versus ORI diagnoses

 

 

 

-- there were overlapping confidence intervals when looking at different times after the index infection between those with Covid and ORIs:

    -- this chart did exclude patients up to one month after the Covid or other viral infections, to limit including the potential hyperglycemic effect of the relatively transient acute infections

    -- of note, there seemed to be an increasing number of Covid diagnoses at all time-frames measured

 including up to nine months after the infection

 

Commentary:

-- diabetes had been found after Covid infection in adults; a systematic review and meta-analysis of eight eligible studies of adults with more than 4 million people having Covid vs more than 43 million controls done between 2020-2022, found a 66% higher risk of incident diabetes in those with Covid: https://www.nature.com/articles/s41598-022-24185-7 ; those studies  found more new onset diabetes in males and in those with more severe Covid infections

-- another large data-mining studies using electronic health records found an increased risk of subsequent type 1 diabetes in children having Covid , and another documented an increase in autoimmune complications including multisystem inflammatory syndrome or myocarditis

-- it should be noted that type II diabetes has been increasing globally in children (from 2001 to 2017, individuals 10 to 19 years old who developed diabetes increased from 0.34 to 0.67/1000 youths, a 95% relative increase), likely related to increased levels of overweight/obesity and insulin resistance

 

-- This current study found that patients who had no indication of having diabetes prior to Covid or ORI had a significantly elevated risk of developing type II diabetes after covid, and this was true at each point they measured (at 1 month, three months, and at six months), as compared to kids with other respiratory infections

 

-- chronic inflammation may serve as a potential physiologic link between covid and diabetes: studies have found that inflammation is associated with insulin resistance and metabolic dysregulation, as precursors of metabolic syndrome and diabetes (https://bmcendocrdisord.biomedcentral.com/articles/10.1186/s12902-021-00925-0), and perhaps the small number of kids (or adults) who develop diabetes after a covid infection are the ones having higher levels of chronic inflammation (see next study on long covid below)

-- also SARS-CoV-2 virus can infect human pancreatic beta cells, perhaps being another mechanism for the association with diabetes

-- some children may also have a genetic susceptibility to have an autoimmune response to SARS-CoV-2 virus leading to anti-beta cell antibodies and potentially type 1 or 2 diabetes

 

-- the researchers argue that even though there were only a few kids who seemed to develop diabetes, this finding is still important because:

    -- kids may well have diabetes for a really long time

    -- the diabetes may well lead to many long term costs, including for medications (increasingly expensive), loss of future health and income (with more costs overall to support them, more absenteeism from future jobs, etc), and a fundamental change in the current and future life of the infected individuals, with much more stress and disruption of the extended family to help provide care to the child (leading to both more financial stress as well as to the many associated medical and social problems associated with stress. As an example, stress is associated with chronic inflammation, and studies have found increased cardiovascular disease: https://gmodestmedblogs.blogspot.com/2022/01/stress-induced-cardiovascular-disease.html )

 

Limitations:

-- this was a retrospective large cohort data-mining study, where the diagnoses and any demographic or medical information were dependent on accurate coding within the database.

    -- as with all associations assessed through data-mining, these results do not imply causality, since many potential contributing factors (psychosocial, medical, genetic...) were not incorporated into this analysis

-- there could be an ascertainment bias: were those kids with covid seeing clinicians more frequently and/or did those clinicians ordering more lab tests (including for diabetes) than those with ORIs? perhaps they did so only in the covid patients, since there had been evidence in adults of an association. so, were those with covid selectively found to have diabetes?

-- there could be reverse causality, where kids with unknown baseline diabetes were predisposed to more serious covid infections that led to uncovering the diabetes? and we know that diabetes seems to make covid worse (and perhaps having more symptomatic vs asymptomatic covid infections), all tending to magnify an association with diabetes?: https://pmc.ncbi.nlm.nih.gov/articles/PMC10378192/#sec4-diagnostics-13-02436

    -- the likelihood of reverse causation in this study is less likely a major factor given that the diabetes diagnoses had a trend to be increased at the 3 month mark vs at the 1-month mark

    -- and, as with other infections, it is possible that for some of the kids who had hyperglycemia, it from the  transient acute respiratory infection, which might last some time (ie, it would have been interesting to have early markers of diabetes at the time of onset of infection and over months of followup)

        -- there is the phenomenon of "glucotoxicity", wherein hyperglycemia itself can lead to insulin resistance, leading to more hyperglycemia. and, resolution of the infection (eg with antibiotics for bacterial ones, time for most viral ones) can improve insulin sensitivity. but some individuals actually require short-term insulin to do so: https://pmc.ncbi.nlm.nih.gov/articles/PMC8050380/

-- it is possible that there was "contamination" in the study, where some/many people with ORIs actually had likely mild covid infections that were not detected perhaps because the symptoms were so minor and the patient/clinician did not test for covid or did not test enough times; this would actually lead to underestimation of the true covid effect on diabetes

 

so, this study does have impressive data (with the above limitations) as well as reasonable physiologic mechanisms that reinforce a likely relationship between covid infections in kids and the development of diabetes.

-- and this does reinforce the social/public health need to still consider covid infections as potentially very severe, even in kids, reinforcing the importance of self-case (masks, minimizing being at superspreader events) and continued vaccinations (a recent analysis found that long covid prevalence, around 7% in adults, is decreased considerably by getting the new covid vaccine; https://gmodestmedblogs.blogspot.com/2024/07/long-covid-prevalence-in-us.html . unfortunately, there seems to be low uptake of this vaccine, at about 22% of adults...

 

-------------------------------------------------------------------------------------------------------------------------------

another recent study of covid-infected individuals, using very advanced and much more sensitive blood assays for SARS-CoV-2 antigens, found an impressive relation of antigen positivity with "long covid", now more frequently referred to as “post-acute sequelae of covid-19” or PASC: this antigenemia could explain at least some cases of long covid (see covid long persistent antigen ClinMicroInf2024 in dropbox or doi: 10.1016/j.cmi.2024.09.001 )

 

Details:

-- 1569 blood samples were analyzed from 706 individuals infected with SARS-CoV-2 virus from 4 independent studies and were analyzed for the S1 subunit of spike, full-length spike or nucleocapsid antigens of the SARS-Cov-2 virus:

    -- Long-term Impact of Infection with Novel Coronavirus at the University of California San Francisco, LIINC (n=171)

    -- Seattle COVID-19 Cohort Study to Evaluate Immune Responses in Persons at Risk  and with SARS-CoV-2 Infection at the Allen Institute (n=80)

    -- Researching COVID-19 to Enhance Recovery (RECOVER), n=392, and the only one with multiple sites (83)

    -- Mass General Brigham in Boston, MGB (n=63)

-- they also assessed 34 commonly reported post-acute sequelae of covid infection (aka, long covid)

 

Results:

-- 706 individuals infected with SARS-CoV-2, 21% (18-24%) were positive for the S1 subunit of spike, full-length spike or nucleocapsid antigens

    -- the full spike antigen was the most commonly detected: 20% (18%-22%), between 4 and 7 months postinfection

        -- the presence of this spike antigen varied dramatically from 5% to 51% in the different studies

    -- but none of the people who had only the acute covid infection (ie, no long covid symptoms) had a detected spike antigen

-- 578 of them had at least one of the long covid symptoms for at least 1 month post-infection (the definition of long covid), most commonly cardiorespiratory, musculoskeletal, and neurologic symptoms in over ½ of the patients

    -- 43% (40% to 45%) of those with long covid were antigen positive

--128 patients (18%) reported no ongoing symptoms but antigen was detected in 28 of them (21%)

 

-- the presence of antigen, controlling for age, sex, time postinfection, and the cohort assessed,  was associated with an 80% increased risk of long covid symptoms, OR 1.8 (1.4-2.2)

    -- antigen positivity across all 4 cohorts was most commonly associated with fatigue, brain fog, muscle pain, joint pain, backpain, headaches, sleep disturbance, smell/taste changes, and gastrointestinal symptoms

    -- in some cohorts, antigen was also detected in some individuals who did not report any ongoing symptoms beyond the post-acute phase, highest in the RECOVER group (36%), least in the MGB group (0%)

 

-- in multivariable logistic regression analysis, female sex, time post-infection, and antigen presence were independently associated with increased odds of having long covid, and most often in the MGB cohort

 

 

Commentary:

-- as with several other infections, there may be prolonged viral presence (both Zika and Ebola viruses, both single-stranded RNA viruses as with SARS-CoV-2, have persistent viral reservoirs)

-- the presence of the antigens varied dramatically from one study to another, especially in the most important spike antigen (see limitations below)

-- there are several potential mechanisms as to why long covid/post-acute sequelae of covid happens:

    -- viral persistence

    -- reactivation of latent virus

    -- autoimmune response   

    -- chronic immunologic dysfunction

-- some other studies have found SARS-CoV-2 RNA in autopsy tissues weeks to months after infection

-- the fact that 28 patients (21% of the 128 patients without long covid symptoms) had antigenemia without symptoms is not surprising:

    -- some individuals are more ”somatic” and more attuned to minor changes in their bodies than others, who are more “deniers”

    -- there may well be different thresholds of what people attribute to covid, depending perhaps even on their political perspective (as we know, even public health is more politically partisan….)

    -- the ascertainment of long covid symptoms varied between the studies, with some being self-reported symptoms, others with more rigorous interviewing by the researchers

 

Limitations:

-- one concern is the variability of results in the antigens between the four different studies (eg it was quite dramatic that the spike antigen was found in 5% in one cohort but 51% in another):

    -- the studies were done at somewhat different times during the pandemic when different SARS-CoV-2 variants were predominant, and there were differences between the variants, in part by the effect of prior immunity and vaccinations

    -- the researchers postulate that this is because of differences in study design, with different inclusion criteria (selection bias), noting that the RECOVER and LIINC studies recruited people during their acute infections, which is very different from recruiting patients who had long covid symptoms

    -- also RECOVER had self-reported symptoms to define long covid, where other studies had more rigorous interviewer-administered symptom surveys

    -- there was also less follow-up information for some of the studies

         -- this issue of study variability, of course, applies to essentially all systematic reviews/meta-analyses that combine different studies: all I have seen over the years have recruited patients from different demographics, have different inclusion/exclusion criteria, different background medications/therapies, etc that undercut the robustness of combining their conclusions.

-- also, some of the cases of long-term spike antigen detection might be from reinfection. the researchers note that the estimated rate of reinfection is 2.5%/year and is unlikely to be a major factor, given the much higher numbers of spike antigen detection in their study

    -- and, they also noted that the spike antigen was detected most often between 4 and 7 months after infection (most commonly around 6 months)

        -- this does not fit in well with prior studies of acute SARS-CoV-2 infections, there are decreasing antigen levels as antibody production increased

        -- this does fit in with other studies finding continued post-acute antigenemia in kids developing multisystem inflammatory syndrome as well as their finding that the presence of spike antigen strongly correlated with symptoms of long covid

    -- would they have found antigenemia more often if they checked for it more frequently? Or by more sensitive assays? (this current assay is much more sensitive than prior ones. But would an even more sensitive one find more correlation with symptoms?) Are there other antigens that are important that were not assessed?

 

so, this study did find a few likely useful things:

    -- viral persistence seems to be a mechanism to long covid for many patients, and the 3 antigens assessed were detected up to 14 months postinfection

    -- the spike antigen seems to be a pretty specific marker for long covid: this study found that the presence of antigens was consistently associated with having at least one long covid symptom

    -- the strong correlation between the long-term prevalence of the spike antigen may well provide an important marker of long covid, perhaps allowing for a more accurate definition of long covid, especially given the large array of long covid symptoms that have potential conflation with other diseases (??is cognitive decline related to long covid? Or sleep disturbances? Or depression? Or stress????)

        -- and, this marker may make it easier to assess future interventions to decrease long covid (especially important, given the evolution of covid into a recurrent endemic infection).  For example, given the association of symptoms with the spike antigenemia, there may be pharmacologic interventions against the spike antigen and (if the relationship found in this study is causal), provide a specific target for intervention.

          -- however, it is important to point out that prolonged spike antigenemia is present in a minority of those with long covid (about 20%):

               -- is that because there are other antigens that were not assessed that are associated with long covid? And we need a broader net in future studies?

               -- is the issue the antigens themselves or is it just the prolonged systemic inflammation created by persistent antigenemia? (this study did not study this possibility, though there have been found long-term antinuclear antibodies in some patients with long covid, or several other inflammatory proteins (eg, see https://www.nature.com/articles/s41467-023-38682-4 ). And chronic inflammation can affect cognitive function: https://pubmed.ncbi.nlm.nih.gov/22743812/

               -- and, as mentioned above, at least some of the patients with “long covid” may well have other conditions and were misclassified as “long covid”

geoff

-----------------------------------

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@bidmc.harvard.edu

  

to get access to all of the blogs:  go to http://gmodestmedblogs.blogspot.com/ to see the blogs in reverse chronological order

or you can just click on the magnifying glass on top right, then type in a name in the search box and get all the blogs with that name in them