key: cord-0962100-xfg3iczd
authors: Zhong, Yajuan; Cao, Yacong; Zhong, Xiaozhu; Peng, Zhihang; Jiang, Sushi; Tang, Tiantian; Chen, Hai; Li, Xiaojia; Xia, Yankai; Cheng, Yanxiang; Zhao, Xiaomiao
title: Immunity and Coagulation/Fibrinolytic Processes may Reduce the Risk of Severe Illness in Pregnant Women with COVID-19
date: 2020-10-22
journal: Am J Obstet Gynecol
DOI: 10.1016/j.ajog.2020.10.032
sha: f05cf922797e40b72f8da6e18de97ed53044da23
doc_id: 962100
cord_uid: xfg3iczd

Background There are specific physiologic features regarding the immunity and coagulation among pregnant women, which may play important roles in the illness development of COVID-19. Objective To determine the key factors associated with the deterioration of patients with COVID-19 and the differentiating clinical characteristics of pregnant women with COVID-19, to interfere with the progression of COVID-19. Study Design A retrospective study of 539 Chinese Han adult patients with COVID-19 was conducted, of which 36 cases were pregnant women. 36 pregnant women without COVID-19 were recruited as the control. The characteristics of severe and critical illness which were differentiated from mild and moderate illness in patients with COVID-19 were analyzed using a machine learning algorithm. Additionally, major differences between pregnant women with COVID-19 and age-matched non-pregnant women with severe/critical COVID-19, paired with pregnant women without COVID-19, were explored to identify specific physiological features of pregnant women with COVID-19. Results For the total patient population, the lymphocyte, CD3+, CD4+, CD8+, CD19+ and CD16+56+ cell counts were significantly lower, and white blood cell (WBC), neutrophil and neutrophil-to-lymphocyte ratio (NLR) were higher in those with severe/critical illness than those with mild/moderate illness (P<0.001). The plasma levels of IL-6, IL-10 and IL-6 to IL-10 ratio (IL-6/10) were significantly increased in critical patients, compared to mild, moderate and severe patients (P<0.001). The above immunological co-clusters achieved an AUC of 0.801 (95% CI: 0.764-0.838); and its combined model with the coagulation and fibrinolysis index (prothrombin time, d-dimer) achieved an AUC of 0.815 (95% CI: 0.779-0.851) using the random Forest regression model to predict severe or critical illness. For the pregnant women with COVID-19, none had pre-existing diseases. They displayed increased WBC, neutrophil count, NLR, and levels of D-dimer and fibrinogen, along with decreased lymphocyte and IL-4 level (P<0.05), compared with non-pregnant women with mild/moderate COVID-19. Although they presented similar changes of immunological markers of lymphocyte, WBC, NLR, CD3+, CD4+, CD8+, CD16+56+ cell count and IL-6/10 compared with non-pregnant women with severe/critical COVID-19, none of the pregnant women with COVID-19 deteriorated into severe or critical illness. There were no significant differences in comparison to WBC, lymphocyte, neutrophil, NLR, immunological markers or coagulation fibrinolysis markers between pregnant women with COVID-19 and pregnant women without COVID-19. As for the discrepancy of pathophysiological features between pregnant women with COVID-19 and non-pregnant women with severe/critical COVID-19, the immunological markers achieved an AUC of 0.875 (95% CI: 0.773-0.977); and its combined model with coagulation and fibrinolysis index achieved an AUC of 0.931 (95% CI: 0.850-1.000). Conclusions Immune dysregulation was identified as a crucial feature of COVID-19 patients which developed severe or critical illness, and pregnant women with COVID-19 presented with similar immune responses but rarer incidences of severe or critical illness. Immune dysregulation is related to the risks of deterioration into severe or critical illness. The specific coagulation/fibrinolysis system of pregnancy may reduce pregnant women with COVID-19 without pre-existing disease from the development of severe illness.

The pandemic of coronavirus disease 2019 (COVID-19) caused by severe acute 86 respiratory syndrome coronavirus 2 (SARS-CoV-2) has infected more than 33.2 million 87 people in most of the countries worldwide, and up to September 29, 2020, 1000040 people 88 have died, as reported by the World Health Organization (WHO) 1 . Despite tremendous efforts 89 to combat this novel infectious disease globally, only a few medications have shown 90 potentially curative effects. The coexistence of SARS-CoV-2 with human beings seems 91 inevitable 2 . Finding a way to interfere with the development of illness to severe or critical 92 stages and reducing the mortality as much as possible may be another strategy 3 . 93 According to the current reports which provide information regarding pregnant patients, 94 the illness severity of pregnant women with COVID-19 is similar with the common 95 population or even lighter 4-10 . In a study conducted in Wuhan city, only 9 of 118 pregnant 96 lymphocytopenia along with an elevated neutrophil-to-lymphocyte ratio (NLR) 16 were 107 reported to be associated with hospitalization rate, severity of illness and/or mortality. On the 108 other hand, dysregulated coagulation and diffuse thrombosis have been observed in lungs and 109 extra-pulmonary organs in deceased patients with acute respiratory distress syndrome 110 (ARDS) 17-19 . 111 However, during pregnancy, the neutrophil, NLR and D-dimer also increase with the 112 progression of gestational age. The microcirculation will undergo modifications during the 113 course of pregnancy to aid adaptation to ensure adequate oxygen supply to the fetus 20 . We 114 hypothesize that the specific immunity response and coagulation-fibrinolysis state among 115 pregnant women COVID-19 may play important roles in progression of illness in Herein, the clinical characteristics and laboratory indicators in 539 Chinese Han 117 COVID-19 patients including 36 pregnant women were analyzed. Vital pathological features 118 distinguishing severe or critical illness from mild or moderate illness were identified via 119 machine learning algorithms. The physiological features between pregnant women with 120 COVID-19 and their counterpart controls were then compared to determine the key factors 121 that may hint the potential mechanism of illness development of COVID-19. 

The clinical, laboratory and outcome data for patients with COVID-19 were obtained 141 from the electronic medical records during the time of admission and discharge for mild 142 illness; admission, improvement and discharge for moderate illness; as well as admission, 143 exacerbation, improvement and discharge for severe or critical illness. Additionally, 144 laboratory data for pregnant women in the control group were collected on the date of 145 admission. The data collection forms were independently reviewed by two researchers. 146

For laboratory assessments, blood samples were collected in the morning after fasting 147 overnight to examine the following indexes: complete blood count (photoelectric colorimetry, 148 XN-9000), coagulation and fibrinolysis index (immunoturbidimetry, CS-5100), N-terminal 149 pro-brain natriuretic peptide (NT-proBNP) (electrochemical laser method, E602), creatine 

The severity of the disease process is determined according to the Diagnosis and 157 Moderate illness is defined as fever, respiratory tract symptoms, and imaging displaying signs 160 of pneumonia. Severe illness is defined as tachypnoea (≥30 breaths/ min) or an oxygen 161 saturation (SaO 2 ) of ≤ 93% on room air at sea level, or a ratio of arterial partial pressure of 162 oxygen to fraction of inspired oxygen (PaO 2 / FiO 2 ) of ≤ 300 mmHg, or progression of 163 lesions >50% within 24-48 hours, as determined by pulmonary imaging. Critical illness is 164 defined as respiratory failure requiring mechanical ventilation, shock or other organ failure 165 that requires intensive care. considered statistically significant. Violin plots were generated with GraphPad Prism version 173 8.0 program (GraphPad Software Inc., San Diego, CA, USA). 174

For the methodology of grouping, firstly, a total of 539 patients with COVID-19 were 175 divided into four sub-groups: mild (n=22), moderate (n=297), severe (n=169) and critical 176 (n=51), according to the severity of illness. Then, 36 pregnant women with COVID-19 and 177 their counterparts of 82 non-pregnant women with COVID-19 aged 20-40 years old which 178 were age-matched were extracted from the total patients. The sub-groups were categorized 179 into pregnant women with COVID-19 (n=36), age-matched non-pregnant women with 180 mild/moderate COVID-19 (n=72) and non-pregnant women with severe/critical COVID-19 181 (n=10) to analyze the biochemical characteristics. In addition, considering the multiple 182 repeated times of blood investigations performed in this study, generalized mixed models and 183 generalized estimating equation models were also used to assess the difference of laboratory 184 characteristics between different sub-groups 22 . These data were analyzed using STATA 185 software (Version 12.0; Stata Corporation). A random Forest (RF) regression model was 186 further used to evaluate the distinguishing abilities of clinical and laboratory markers. The 187 data set was randomly split into a training set containing 80% of the samples, and a validation 188 set containing the remaining 20%. Variables used for distinguishing were selected based on 189 comprehensive consideration of ranking of importance ( Figure S1 ). The RF was 190 implemented using R software version 3.2.2 (R Core Team 2014) 23 using the "random Forest" 191 package. The graphical ROC (Receiver Operating Characteristic) curve was produced along 192 with area under the ROC curve (AUC). Statistical significance was defined as p<0. 05. 193 Results 194

The clinical and biochemical features of all 539 patients were first analyzed, as shown in 196 Table  207 S1. 208

On admission, lymphopenia was detected in 251 (46.6%) patients with COVID-19, as 210 shown in Table S1 . The lymphocyte count, CD3 + , CD4 + , CD8 + , CD19 + and CD16 + 56 + cell 211 counts were significantly lower in severe and critical patients compared with mild and 212 moderate patients (P<0.001). Serum white blood cell (WBC), neutrophil count and NLR were 213 significantly higher in severe and critical patients than moderate patients (P<0.001), as shown 214

in Table S1 . Corresponding with the cellular immune dynamic in the aggravating severity of 215 (IL-6/10) were significantly increased in critical patients, compared to the mild, moderate and 217 severe patients (P<0.001), as shown in Table S1 . However, it is noteworthy that the incidence 218 of more than triple the normal cut-off value of IL-6 (>60.0 pg/mL) was 34.6% (9/26) in 219 critical patients and 1.3% (1/78) in severe patients. No significant differences were found in 220 the serum levels of the other cytokine co-cluster (i.e. IL-2, IL-4, tumor necrosing factor 221 (TNF)-α, and interferon (IFN)-γ) among the four sub-groups. 

The median age of the 36 pregnant women with COVID-19 was 29.0 years old (IQR 243 27.0-32.0). The gestational age was from 5 weeks to 41 weeks and 69.4% women were in the 244 3 rd trimester of pregnancy. 19.4% ( weeks. No individual in control group had comorbidity ( Table 2) . 251 Pregnant women with COVID-19 displayed similar changes of immunological markers 252 of WBC, lymphocyte, WBC, NLR, CD3 + , CD4 + , CD8 + , CD16 + 56 + cell count and IL-6/10 253 compared with non-pregnant women with severe/critical COVID-19 (Figure 2A and 2B) , and 254 had increased WBC, neutrophil count, NLR, and levels of D-dimer and FIB, as well as 255 decreased lymphocyte and IL-4 level (P<0.05), compared with non-pregnant women with 256 mild/moderate COVID-19 (Table 3 and Table S6 ). 257 non-pregnant women with severe/critical COVID-19, there are significant difference in 267 immunological and coagulation/fibrinolysis indicators. 268 

In this retrospective study with the clinical outcomes of COVID-19 patients, the immune 287 dysregulation was identified as a vital feature of COVID-19 patients which developed severe 288 or critical illness, and pregnant women with COVID-19 presented similar immune response 289 with non-pregnant women with severe/critical COVID-19 but rare incidence of severe or 290 critical illness. Digging deeper into the discrepancy between the pregnant women with 291 COVID-19 and non-pregnant women with severe/critical COVID-19, the coagulation 292 fibrinolysis index (PT, D-dimer) showed remarkable differences. However, for pregnancy 293 itself, the pregnant women without COVID-19 also demonstrated similar changes of 294 immunological markers, consisting of WBC, lymphocyte, neutrophil and NLR, as the 295 pregnant women with COVID-19. 296

In our study, pregnant women with COVID-19 presented with milder symptoms and 298 mild illness, compared to the non-pregnant women with COVID-19 of reproductive age. The 299 result is partly consistent with a meta-analysis including 13118 pregnant and 83486 300 non-pregnant women with COVID-19 which demonstrated that pregnant women with 301 COVID-19 had milder symptoms. Although the meta-analysis indicated that pregnant women 302 with COVID-19 were more likely to require admission to an intensive care unit (ICU) (1.62, 303

1.33 to 1.96; I 2 =0%) at first glance, it was further elucidated that severe COVID-19 or 304 requirement of admission to an ICU were actually associated with older age, obesity and 305 pre-existing maternal comorbidities like hypertension, diabetes, etc 24 . Our data also indicated 306 that risks of severe and critical COVID-19 were associated with the above factors, in 307 accordance with these reports on the other hand. The majority of the pregnant patients in our 308 study had no comorbidities or complications, which may partly explain their mild to moderate 309 illness. The maternal immune system is unique during normal pregnancy. However, to date, In general, normal maternal immune system is actively modulated at different gestational 314 stages. The first trimester is a pro-inflammatory state in favor of embryo implantation, the an anti-inflammatory state benefiting fetal growth, and the third trimester changes to a second 317 pro-inflammatory state featured by a Th1-type immune environment necessary for labor. Viral 318 invasion switches the immune environment from the Th2-type to Th1-type and activates 319 inflammation by overproduction of cytokines including IL-6 10 . The pregnant women with 320 COVID-19 in the third trimester composed nearly 70% of the whole pregnant population in 321 our study. This may explain why the immunological files are similar between pregnant 322 women with COVID-19 and pregnant women without Our data indicated that for the general COVID-19 women population, the immune 324 dysregulation mainly manifesting as increased IL-6 , IL-10 and IL-6/10 levels and a decreased 325 lymphocyte count and its subsets, along with increased NLR, is related to the risks for severe 326 and critical illness. The pregnant women with COVID-19 demonstrated similar immune 327 changes but none severe illness occurred. The most dramatic discrepancy between pregnant 328 women with COVID-19 from non-pregnant women with COVID-19 was the coagulation 329 fibrinolysis index variance. 330

The significant increase of IL-6 (pro-inflammatory cytokine), IL-10 (an anti-inflammatory 331 cytokine) and IL-6/10 ratio in severe or critical illness indicated that the host immune system 332 switches to a predominantly pro-inflammatory state. It was reported that infection with 333 SARS-CoV-2 triggers a pro-inflammatory response characterized by the production of IL-6, 334 C-X-C motif chemokine 10 (CXCL10) and type 1 interferons, which attract macrophages, 335 monocytes, and T lymphocytes to infection sites 26 . However, in our study, there were no 336 significant differences in terms of serum IL-2, IL-4, TNF-α, and IFN-γ levels within 337 comparisons among defined sub-groups. It demonstrates the heterogeneity of the host 338 immune response to SARS-CoV-2. 339

It is notable that pro-inflammatory cytokines (mainly IL-6) are common triggers to Besides, lymphopenia as a factor related to severe COVID-19 illness has been 358 reported 32-34 . Our data indicated that the absolute counts of total lymphocyte and its subsets 359 were decreased and sustained in severe and critical patients, along with increased NLR, 360 predicting the risks for severe or critical COVID-19 illness, in accordance with Qin's study 35 361 that lymphopenia, especially reduction of CD4 + and CD8 + T cells, was related with 362 deterioration and signs of poor prognosis 36, 37 . Additionally, neutrophil extracellular traps 363 Therefore, immunity disorders mainly consisting of increased IL-6, neutrophil and NLR, may 368 trigger thrombotic events by extrinsic and intrinsic pathways simultaneously, promoting the pathological process to severe COVID-19. 370

The main finding of our study is the association between excessive immune response, 372 coagulation-fibrinolysis dysregulation, and severity of the disease in both general population 373 and pregnant sub-group. Given the unique immune state in pregnancy, as described before, a 374 deeper investigation need to be performed to figure out the exact mechanism between 375 SARS-CoV-2-induced inflammation and coagulopathy. In addition, in a series of 11 autopsy 376 cases of patients with severe COVID-19 from Australia, all cases were found pulmonary 377 arterial obstruction by thrombotic material at both the macroscopic and microscopic levels. 378

Interestingly, ten of these cases received pharmacologic venous thrombus embolism (VTE) 379 prophylaxis, and VTE was not clinically suspected in any cases before autopsy as a 380 contributor of death 18 . A method to evaluate microcirculation disorders and identify early alert 381 biomarkers is needed. Early anticoagulant therapy is implicated according to our study. vasodilating, pro-secretory, and anti-inflammatory effects, which might be assumed to be 387 potential protective mechanisms against the physiological process of COVID-19 and need to 388 be further evidential. Furthermore, the presence of pregnancy may lead to adaptive changes in 389 maternal immunity to accept the existence of the fetus as a semi-alien. Likewise, it is also 390 possible that there is an unknown mechanism temporarily in mater compromising the 391 invasion of SARS-CoV-2. Therefore, long-term follow-up of pregnant women with 392 COVID-19 and their offspring is necessary. Besides, we are also concerned about 393 microcirculatory impairments related to endothelialitis and microangiopathy, since some 394 adverse pregnant outcomes, e.g. preeclampsia and HELLP syndrome, are caused by this 395 physiological change 42 . 396

The strengths of the study include that using machine learning algorithm to clinical data 398 sets for the purpose of developing robust risk models of COVID-19; and compared pregnant 399 women with COVID-19 and age-matched non-pregnant women with severe/critical 

The 36 pregnant women with COVID-19 and 82 non-pregnant women with COVID-19 (A) or 574 10 non-pregnant women with severe/critical COVID-19 (B) were extracted for this analysis. 575 ROC curves for immunological markers of NLR, CD3 + , CD4 + , CD8 + , CD16 + 56 + cell count, 576 IL-6, IL-6/10, lymphocyte count, and WBC (a); immunological markers and LDH (b); J o u r n a l P r e -p r o o f J o u r n a l P r e -p r o o f 

MMWR Morbidity and mortality weekly report

Report of the WHO-China Joint Mission on Coronavirus Disease

An inflammatory cytokine signature predicts 461 COVID-19 severity and survival

Epidemiological and clinical characteristics of 99 cases of 2019 463 novel coronavirus pneumonia in Wuhan, China: a descriptive study. The Lancet

Clinical course and risk factors for mortality of adult inpatients with 466

China: a retrospective cohort study. The Lancet

Clinical Characteristics of Coronavirus Disease 2019 in China

Pulmonary Arterial Thrombosis in COVID-19 With Fatal Outcome

Clinicopathologic Case Series

Autopsy Findings in 32 Patients with

Single-Institution Experience

Microcirculation in pregnancy. Physiological 478 research

Team R. A language and environment for statistical computing

Author Contributions: D. Zhao and Cheng had full access to all of the data in the study and