Spontaneous Hypothermia As an Indicator of Early Diffuse Anoxic Brain Injury in Post-Cardiac Arrest Patients

Abstract

OBJECTIVES:

To determine the association between spontaneous hypothermia (SH), defined as initial post-resuscitation core body temperature less than 34°C, and diffuse anoxic brain injury (DABI) on initial CT scan of the head (CTH) in post-cardiac arrest patients.

DESIGN, SETTING, AND PARTICIPANTS:

This was a retrospective, observational cohort study. This study was performed at the University of Rochester Medical Center Strong Memorial Hospital. All in-hospital and out-of-hospital cardiac arrest patients with return of spontaneous circulation admitted between January 1, 2022, and October 31, 2022, were included.

MAIN OUTCOMES AND MEASURES:

The primary outcomes were the odds of DABI on initial CTH for patients with SH compared with patients without SH post-cardiac arrest using a multivariable logistic regression controlling for patient covariates including basic demographics and arrest features. DABI on initial CTH was measured qualitatively and quantitatively using neuroradiologist interpretation and calculated gray-white matter ratio of the basal ganglia, respectively. Secondary outcome measures included length of stay (LOS), inpatient mortality, and those who underwent withdrawal of life-sustaining therapy (WOLST) or progression to brain death.

RESULTS:

Out of the observed 150 cases of cardiac arrest, 31 patients (21%) had SH. Of the 128 patients who had an initial CTH performed, 27 (21%) had DABI. The adjusted odds ratio of DABI on initial CTH associated with SH was 3.55 (95% CI, 1.08–11.64; p = 0.036) and 2.18 (95% CI, 0.69–6.91; p = 0.182) when DABI was measured qualitatively and quantitatively, respectively, after controlling for multiple covariates. There was a difference observed in LOS between the groups (3 vs. 10 d; p = 0.0005) and this was driven by early WOLST.

CONCLUSIONS AND REVELANCE:

Patients presenting with SH after cardiac arrest may be at greater risk of early DABI on initial CTH compared with those with higher body temperatures in the post-arrest period. Recognition of early SH may help to risk stratify post-cardiac arrest patients at highest risk of DABI.

KEY POINTS

Question: We hypothesize that spontaneous hypothermia (SH), defined as an initial body temperature less than or equal to 34°C, is associated with diffuse anoxic brain injury (DABI) on CT imaging in post-cardiac arrest patients regardless of time to return of spontaneous circulation.

Findings: In this retrospective cohort study, we observed an increased, adjusted odds for DABI in patients with SH of 3.55 (95% CI, 1.08, 11.64; p = 0.036) and 2.18 (95% CI, 0.69–6.91; p = 0.182) when DABI was measured qualitatively and quantitatively, respectively, after controlling for multiple covariates.

Meaning: SH immediate post-arrest may be a potentially useful risk stratification tool.

In the United States, over 500,000 people will suffer cardiac arrest each year (1). Of these, only about 15–22% will survive (1, 2). Irreversible hypoxic-ischemic encephalopathy (HIE) as part of the post-cardiac arrest syndrome and subsequent withdrawal of life-sustaining therapy (WOLST) make up a large portion of the deaths following return of spontaneous circulation (ROSC) (2, 3). The components of diffuse anoxic brain injury (DABI) seen radiographically involve primary injury from cessation of blood flow and secondary injury after reperfusion, resulting in diffuse neuronal death and inflammation (3). Severe DABI can be seen on a CT scan of the head (CTH) as reduction in the gray-white matter ratio (GWR) differentiation and diffuse cerebral edema (4). This manifests clinically as HIE.

The calculated GWR on unenhanced CTH has previously been reported to predict poor neurologic outcome as a result of severe HIE with high specificity in patients who have suffered cardiac arrest (5). Specifically, GWR measured from the basal ganglia (GWR_BG) has demonstrated the highest predictive performance (5–8).

Neuroprognostication after cardiac arrest is a multimodal approach to determine an individual’s severity of HIE and, subsequently, predict their functional outcome (4). Current literature supports utilizing clinical, laboratory, radiologic, and/or electrographic data obtained in a delayed fashion (72–120 hr post-ROSC) for this purpose in comatose post-arrest patients (9, 10). The clinical examination is an easy, low-cost prognostic element that can be followed throughout this period.

Spontaneous hypothermia (SH) in post-cardiac arrest patients at admission has been proposed as an early clinical sign of severe HIE (11, 12). While hypothermia during and immediately following arrest is relatively common in out-of-hospital cardiac arrest (OHCA), core body temperatures closer to normothermia on ICU admission in these patients has been associated with survival to hospital discharge (13). Anoxic damage to the hypothalamus, the center for thermoregulation in the brain, is suggested as the pathophysiology for this phenomenon (11, 12, 14).

We aim to evaluate the association between SH, defined as core body temperature less than or equal to 34°C, with DABI on initial CTH, defined both qualitatively by a neuroradiologist and quantitatively through use of separately calculated GWR_BG.

MATERIALS AND METHODS

Design, Setting, and Study Population

This was a retrospective, single-center observational cohort study performed at the University of Rochester Medical Center in Rochester, NY, which is an 886-bed tertiary care hospital. All adult (age ≥ 18 yr old) cardiac arrest patients with ROSC admitted between January 1, 2022, and October 31, 2022, were included in the study. The study (originally titled “STUDY00007843: Precision of neuroprognostication in cardiac arrest patients”) was approved by our institutional review board (IRB) with a waiver of informed consent. All procedures were followed in accordance with the ethical standards outlined by the IRB and with the Helsinki Declaration of 1975.

Data Collection

All data were obtained through chart review of the institution’s electronic medical record. Research Electronic Data Capture was used to collect and store data. Collected data included de-identified demographics, number of comorbidities, clinical features of the cardiac arrest, results of initial CTH as stated by the on-call neuroradiologist, GWR_BG calculated from the initial CTH by an independent reviewer (S.V.P.), and other outcome measures. Specifically, demographic data included age, self-identified race and ethnicity, and sex. The clinical features of the cardiac arrest included initial cardiac rhythm, time to ROSC (min), the location of arrest (in-hospital cardiac arrest [IHCA] and OHCA), and initial recorded core body temperature. Initial rhythm was dichotomized to shockable (ventricular tachycardia or ventricular fibrillation) and nonshockable rhythm (asystole or pulseless electrical activity [PEA]).

Exposure

Initial post-arrest core body temperature as obtained in the emergency department on patient arrival was dichotomized to greater than 34°C or less than or equal to 34°C. Patients with initial recorded core body temperatures less than or equal to 34°C were defined as having SH and acted as the exposed group. To determine the cutoff for SH, we performed a sensitivity analysis investigating odds ratios (ORs) at various core body temperatures. The strongest association between SH and DABI on CTH upon this initial analysis was observed at temperatures less than or equal to 34°C.

Outcome Measures

The primary outcome measure was presence of DABI on the initial CTH, collected both qualitatively and quantitatively. At our institution, it is standard to perform the initial CTH within 24 hours of arrest if the patient is clinically stable and it does not interfere with post-arrest targeted temperature management (TTM). Results of the initial CTH as interpreted by the on-call neuroradiologist were used as the qualitative primary outcome. These results were retrospectively collected as either the presence or absence of DABI by observation of any decreased gray matter attenuation, decreased GWR differentiation, and/or diffuse cerebral edema. This was rereviewed by the attending neurointensivist at the time of patient care and, again, by a separate neurologist for the purposes of this study (S.V.P.). We calculated GWR_BG on each available initial CTH as the quantitative primary outcome. Hounsfield units of regions of interest in the putamen (PU) and caudate nucleus (CN) were compared with the corpus callosum (CC) and posterior limb of the internal capsule (PLIC) using an equation used in previous studies: GWR_BG = (PU + CN)/(CC + PLIC) (Supplemental Material A, http://links.lww.com/CCX/B320) (6–8). A cutoff of GWR_BG less than 1.18 was used as the threshold for DABI, as previously established with high predictive performance in a previous study comparing various methods and thresholds for GWRs (6).

Secondary outcome measures included length of stay (LOS) for all patients, LOS for the subset of patients who underwent WOLST, inpatient mortality, number of patients who underwent WOLST, and number of patients who progressed to brain death.

Statistical Methods

Differences in patient characteristics and outcomes by presence of SH were compared using chi-square tests for categorical variables and Wilcoxon rank-sum tests for continuous variables. We developed a hierarchical logistic regression analyzing the primary outcome first as a qualitative measure of the presence of DABI per the neuroradiologist interpretation. Model 1 represented the unadjusted OR for patients with SH (exposed) compared with patients without SH (control). From that, we built on this model by utilizing a multivariable logistic regression to calculate an adjusted OR controlling for demographic features and number of comorbidities (model 2), and demographic features, number of comorbidities, and clinical features of the CA (model 3). This same hierarchical logistic regression was then used to analyze the primary outcome as a quantitative measure of the presence of DABI using the GWR_BG less than 1.18 (models 1a, 2a, and 3a). Patients without an initial CTH were excluded from these logistic regression analyses. Statistical significance was set a priori at p value of less than 0.05. Stata Version 18.0 was used for all data analyses (StataCorp, College Station, TX).

RESULTS

From January 1, 2022, to October 31, 2022, there were 150 cardiac arrest cases from 147 patients. Of these, 58 (39%) were female with median age 60 (IQR, 48–71). Mean time to ROSC was 22 minutes (sd = 1.74 min), and the most frequent presenting rhythm was asystole or PEA (71%). Thirty-one patients (21%) had SH. The median time to initial CTH was 3 hours from admission (IQR, 1–6). Of the 128 patients who had an initial CTH performed, 27 (21%) had DABI.

There were no significant differences in the demographics, number of comorbidities, and clinical features of cardiac arrest between patients with SH and those without SH (Table 1).

TABLE 1.

Patient Demographics and Features of Cardiac Arrest by Spontaneous Hypothermia

Patient Characteristics	No-SH (n = 119)	SH (n = 31)
Age, median (IQR)	61 (50–72)	58 (40–70)
Sex, female (%)	44 (37.0)	14 (45.2)
Race, n (%)
Asian	3 (2.7)	0 (0)
Black	30 (27.3)	3 (11.5)
Hispanic/Latino	2 (1.8)	2 (7.7)
White	73 (66.4)	20 (76.9)
Other	2 (1.8)	1 (3.9)
Missing	9 (7.6)	5 (16.1)
Number of comorbidities, median (IQR)	4 (3–6)	3 (1–6)
In-hospital cardiac arrest, n (%)	20 (16.8)	1 (2.7)
Initial rhythm, n (%)
Nonshockable	68 (70.1)	22 (78.6)
Shockable	29 (29.9)	6 (21.4)
Unknown	22 (18.5)	3 (9.7)
Time to return of spontaneous circulationa, median (IQR)	15 (9–25)	22 (9–42)
Time to initial CTHb, median (IQR)	3.1 (1.6–7.2)	2.7 (0.2–4.9)
Diffuse anoxic brain injury on initial CTH, n (%)	15 (12.6)	12 (38.7)

CTH = CT of the head, IQR = interquartile range, SH = spontaneous hypothermia.

^aMissing values exist for time to return of spontaneous circulation (no-SH n = 10, SH n = 7).

^bCTH not obtained for 20 patients (no-SH=18, SH n = 2).

The unadjusted OR for the presence of DABI on initial CTH evaluated separately as a qualitative (model 1) and quantitative (model 1a) measure in patients with SH was 3.95 (95% CI, 1.57–9.93; p = 0.003) and 3.18 (95% CI, 1.36–7.45; p = 0.008), respectively (Table 2). The adjusted OR for the presence of DABI on initial CTH measured qualitatively and quantitatively in patients with SH controlling for age, sex, race and ethnicity, and number of comorbidities was 3.96 (95% CI, 1.53–10.25; p = 0.005) and 3.27 (95% CI, 1.29–8.30; p = 0.013), respectively (Table 2). The adjusted OR for the presence of DABI on initial CTH measured qualitatively and quantitatively in patients with SH controlling for age, sex, race and ethnicity, number of comorbidities, initial rhythm, and time to ROSC was 3.55 (95% CI, 1.08–11.63; p = 0.036) and 2.18 (95% CI, 0.69–6.91; p = 0.182), respectively (Table 2).

TABLE 2.

Primary Outcome by Statistical Model

Statistical Modela	OR (95% CI)	p
1	3.95 (1.57–9.93)	0.003
1a	3.18 (1.36–7.45)	0.008
2	3.96 (1.53–10.25)	0.005
2a	3.27 (1.29–8.30)	0.013
3	3.55 (1.08–11.63)	0.036
3a	2.18 (0.69–6.91)	0.182

OR = odds ratio.

^aStatistical models developed using a hierarchical logistic regression for unadjusted (1) and adjusted ORs of diffuse anoxic injury as identified by neuroradiologist on CT of the head performed within 24 hr of arrest in presence of spontaneous hypothermia (SH) controlling for demographic features and number of comorbidities (2), and demographic features, number of comorbidities, and clinical features of the cardiac arrest (3). Model subsets represent the same statistical models for unadjusted (1a) and adjusted (2a and 3a) ORs of gray-white matter ratio < 1.18 in presence of SH controlling for aforementioned features.

A comparison of secondary outcome measures is summarized in Table 3. There was a difference in median hospital LOS between patients who had SH and those who did not (3 vs. 10 d; p = 0.0005). There was also a significant difference observed in median hospital LOS in the subset of patients who underwent WOLST between the groups (3 vs. 9 d; p = 0.0015). There were no differences observed between SH and no-SH in inpatient mortality (83.9% vs. 68.9%; p = 0.098), WOLST (58.1% vs. 54.6%; p = 0.399), or progression to brain death (16.1% vs. 8.0%; p = 0.275) between the groups.

TABLE 3.

Secondary Outcomes by Spontaneous Hypothermia After Cardiac Arrest

Secondary Outcome	No-SH (n = 119)	SH (n = 31)	p
LOS, median (IQR)	10 (4–34)	3 (1–9)	0.0005
LOS for WOLST, median (IQR)	9 (4–32)	3 (1–6)	0.0015
Inpatient mortality, n (%)	82 (68.9)	26 (83.9)	0.098
WOLST, n (%)	65 (54.6)	18 (58.1)	0.399
Brain death, n (%)	9 (8.0)	5 (16.1)	0.275

IQR = interquartile range, LOS = length of stay, SH = spontaneous hypothermia, WOLST = withdrawal of life-sustaining therapy.

When comparing our qualitative and quantitative evaluation of the primary outcome, a median GWR_BG of 1.12 (IQR, 1.09–1.17) was observed when our clinicians (neuroradiologist, neurointensivist, and neurologist) interpreted DABI on the initial CTH. When GWR_BG threshold for DABI is set at 1.18, our clinicians agree 69% of the time. This increases to 77% of the time when GWR_BG threshold for DABI is decreased to 1.16.

DISCUSSION

SH in post-cardiac arrest patients at admission has been proposed as an early clinical sign of poor neurologic outcome as a result of severe HIE, and anoxic damage to the hypothalamus is the suggested pathophysiology (9, 11, 12, 14). In these studies, the definition of SH was centered around 35°C, and there was no attempt to correlate this clinical finding with a more objective measure as part of the post-cardiac arrest workup. A sensitivity analyses performed to determine the appropriate threshold to define SH in our study did not see a strong association at body temperature of 35°C. In a more recent study, low body temperature was defined as less than or equal to 35.9°C and was not associated with worse neurologic outcome (15). The authors of this study used an even higher threshold than those previous studies. Furthermore, the body temperatures recorded in this study included axillary and ear temperatures, which are not true core body temperature measurements, and may have overestimated the number of patients in the lower body temperature group who were truly normothermic (15). Finally, some of these studies did not control for time to ROSC and other features of the cardiac arrest event that are known to contribute to mortality (11, 15–17).

In our study, we have observed a relationship between SH and the presence of DABI on CTH performed within 24 hours in post-cardiac arrest patients controlling for multiple variables. When DABI was measured quantitatively using GWR_BG, it is important to note that adjusting for initial cardiac rhythm and time to ROSC nullifies the significance of this association. This is not an unexpected finding, as these factors are known predictors of outcome in post-cardiac arrest patients (16, 17). Additionally, previous studies have suggested that GWR is misleadingly higher when CTH is performed within 6 hours of arrest and will be more accurate beyond this timeframe (18). Our median time to CTH was 3 hours. Another notable finding was the difference we observed between the qualitative measure of DABI on CTH using our clinician interpretation vs. the quantitative use of GWR_BG at a predefined threshold of 1.18. Currently, there is no consensus on the appropriate GWR threshold to predict anoxic brain injury post-cardiac arrest (6). Our data suggests our clinicians required less gray-white differentiation to interpret findings suggestive of DABI, an appropriately cautious approach this early on in the post-cardiac arrest period.

We also observed a shorter hospital LOS for patients with SH. This association was driven by early WOLST. This could be due to the fact that patients who present with SH have greater presenting illness severity than those who are normothermic at admission. We did not observe a difference in other secondary outcome measures including inpatient mortality despite clear association of SH with DABI on CTH performed within 24 hours of arrest, which is seemingly paradoxical. We believe this highlights SH as a possible indicator of “early” DABI and can help risk-stratify patients who may have already endured significant brain injury. This may have implications for selecting appropriate candidates for TTM in the hospital setting and as part of clinical trials. For this reason, we support prioritizing core body temperature measurement as early as possible in the post-cardiac arrest period and close monitoring of its trend, a suggestion previously supported by another study that monitored core body temperature trends as a prognostic tool post-cardiac arrest (12). Additionally, we believe timely CTH within the first 24 hours of presentation to the hospital, if clinically feasible, is a vital component of building a complete understanding of the possible extent of HIE for patients presenting with SH.

Neuroprognostication in post-cardiac arrest patients is a difficult task and requires a multimodal approach involving key clinical, laboratory, radiological, and electrographic elements (4). There is currently no standardized process for determining ultimate functional outcome in comatose post-cardiac arrest patients (10). By identifying SH as being associated with early, severe brain injury, we continue to build on the components of the post-cardiac arrest syndrome. Ultimately, with a clearer definition of this condition, clinicians can understand the subpopulation of patients at highest risk for poor neurologic outcome and prioritize aggressive care more appropriately.

Our study is limited due to its retrospective design using a preexisting dataset that was not collected with the purposes of this study in mind, which prevents control of various factors. Our study population is from a single institution and lacked diversity in racial and ethnic groups, limiting the generalizability of this study. The sample size was also small, which limits the ability to perform more complex statistical analyses and, therefore, also limits the strength of conclusions drawn from the data obtained. The time to measurement of initial core body temperature from ROSC was unknown; however, the standard of care within our institution is to obtain a core body temperature within one hour of ROSC. Second, body temperatures are also subject to the environmental conditions of the cardiac arrest location. For example, a patient that suffers a cardiac arrest outdoors in our study location of Rochester, NY may be subject to temperature extremes dependent on the weather, particularly in winter. However, it is important to note that two previous studies did not find higher numbers of OHCA patients presenting with SH in winter months, one of which with similar winter temperatures to our region (12, 15). Finally, the reversibility of DABI and core body temperature trends were not measured as part of this study, which could be an interesting direction for future research.

CONCLUSIONS

An association exists between SH, defined in our study as initial core body temperature less than or equal to 34°C, and presence of early DABI on initial CTH in post-cardiac arrest patients. Prompt identification of early brain injury in this patient population can help risk-stratify patients for aggressive TTM and future clinical trials.