Serum Lactate, an Independent Prognostic Marker in Normotensive Patients With Acute Pulmonary Thromboembolism

We evaluated all patients diagnosed with APE confirmed by contrast-enhanced chest CT scanning who were admitted consecutively to our clinic between January 1, 2014, and December 31, 2021. The exclusion criteria were: being aged under 18 years, lack of CT angiography evaluation, lack of arterial serum lactate measurement on admission, and the presence of hemodynamic instability.

The study protocol was reviewed and approved by the Ethics Committee of “Sf. Pantelimon” Emergency Hospital (license number 49/2018). At the time of admission, patients signed a consent form agreeing that their medical information could be used anonymously for research purposes.

Hemodynamic instability was defined as: cardiac arrest, obstructive shock (systolic blood pressure (BP) <90 mmHg or vasopressor support required to maintain a BP ≥ 90 mmHg associated with signs of hypoperfusion), or persistent hypotension (systolic BP <90 mmHg or a systolic drop in BP ≥40 mmHg for >15 minutes, not caused by new-onset arrhythmias, hypovolemia or sepsis).

The patients were divided into two groups depending on the location of the thrombus on the thoracic CT examination: either central (thrombus present above or at the lobar arteries level) or distal (segmental or subsegmental) APE.

The serum lactate level was determined from arterial blood collected at the time of evaluation in the emergency room and expressed in mg/dL. Arterial blood gas (ABG) was collected, according to the Emergency Department's protocol, from all patients with dyspnea at mild effort or at rest, tachypnea, or O2 saturation in room air < 95%. An elevated level of serum lactate was considered above a value of 18 mg/dL (2 mmol/L). All other biological samples were collected from venous blood and were processed in the central laboratory, including troponin I. A value above 0.2 ng/dL of serum troponin I was considered elevated and indicative of myocardial injury.

Transthoracic echocardiography was performed at the time of hospitalization by a team of certified doctors. RV dysfunction was defined as at least one of the following parameters: –

RV dilatation in parasternal long axis view,

In-hospital all-cause mortality was the primary endpoint. Length of hospital stay was the secondary endpoint.

Data analysis was performed using IBM SPSS Statistics version 25 and EpiInfo 2007.

Quantitative variables were tested for normal distribution using the Shapiro-Wilk Test and expressed as means with standard deviations in the case of Gaussian distribution, or as medians with interquartile ranges otherwise. Qualitative variables were expressed as absolute counts and percentages.

Quantitative variables were tested using Student T/Mann-Whitney U tests according to their distribution. Qualitative variables were tested using a two-tailed p Fisher's Exact test. Odds ratios with 95% confidence intervals were used to illustrate the measure of associations observed in the contingency tables.

ROC analysis was employed for determining the correlation between numerical and categorical variables. Youden's index-associated criterion was used to establish the optimal cut-off values for lactate levels in the prediction of in-hospital all-cause mortality.

Multivariable logistical regression, specifically the Enter method, was used to identify independent predictors of in-hospital mortality, using the variables identified to be correlated with the outcome in univariable analysis. Prediction models were derived from the multivariable analysis.

161 normotensive APE patients with serum lactate levels determined on admission were included in the study from the initial cohort of 255 subjects (Figure 1).

Figure 1

The mean age was 68.61 ± 11.54 years. 54.94% were female. The clinical and paraclinical characteristics are represented in Table 1. In-hospital mortality was 19.88%. The median duration of hospitalization was 10.5 [7–52] days.

44.7% of patients had serum lactate levels above the upper limit of normal (> 2 mmol/L /18 mg/dL). Their hospital stay durations were similar to those of patients with normal lactate levels (p = 0.43).

In-hospital mortality was higher in patients with elevated lactate levels (27.78% versus 13.48%, p = 0.029). Except for a higher prevalence of deep vein thrombosis (DVT) (54.10% versus 36.25%, p = 0.04), the clinical parameters were similar among patients with increased versus normal lactate levels. Of the laboratory biomarkers, patients with higher lactate levels had increased troponin I (0.379 [0.2–1.65] versus 0.215 [0.2–1.007] ng/mL, p = 0.051), increased NT-proBNP (3704 [1405.5–9959.5] versus 1119.5 [322.2–4960.7] pg/mL, p = 0.002), and increased creatinine levels (1.05 [0.87–1.28] versus 0.98 [0.75–1.13] mg/dL, p = 0.042). These patients associated more frequently right ventricular dysfunction (63.89% versus 36.11%, p = 0.044), as well as the S1Q3T3 sign on ECG (50.00% versus 27.27%, p = 0.004).

PESI score was the main driver of in-hospital mortality (Table 1).

In ROC analysis, serum lactate was a predictor of in-hospital mortality with an AUC of 0.662 (95%CI 0.584 – 0.735, p = 0.005). The cut-off level identified by the Youden index-associated criterion was > 38 mg/dL with a 34.38% sensitivity and 94.57% specificity (Figure 2).

Figure 2

Higher lactate levels > 38 mg/dL were associated with higher PESI scores, hypoxia, the presence of a right heart thrombus, ST elevation in precordial leads, and the need for inotropic support (Table 2).

The lactate cut-off > 38mg/dL (OR 9.12, 95% CI 3.17 – 26.21, p < 0.001) had a better predictive power for in-hospital mortality compared to the upper limit of normal > 18mg/dL (OR 2.42, 95%CI 1.09 – 5.37, p = 0.04).

In multivariable analysis for in-hospital mortality alongside the PESI score, the lactate level evaluated as a continuous variable was an independent predictor for the outcome, with a 5% increase in risk of fatality with each 1 mg/dL increase in lactate value (Table 3). When the lactate level was assessed as a dichotomous variable, the derived limit > 38mg/dL was associated with an independent HR for mortality of 7.49 (95%CI 2.41 – 23.25), while the limit > 18mg/dL lost statistical significance (Table 3).

The predictive model 3, derived from the multivariable analysis that included the PESI score and the lactate levels > 38 mg/dL with an AUC of 0.769, outperformed the predictive value of the PESI score alone for in-hospital mortality (AUC 0.725), as well as the predictive value of the 2019 ESC guidelines algorithm (PESI score, RV dysfunction and troponin - AUC 0.736). Adding lactate levels > 38 mg/dL to the ESC proposed algorithm rendered the optimal prediction power (AUC 0.807) (Table 4, Figure 3).

Figure 3

We acknowledge some limitations to our study, particularly the retrospective application of the scores as well as the relatively small number of patients included in a single center. The included patients were diagnosed and treated over a lengthy period of time, which may have limited the reliability of the present results. Another limitation is that the plasma lactate level of each patient was measured only once. It is likely that repeated measurements would result in improved prognostic information.

Higher mortality rates were observed in our cohort compared to previously reported data from large APE registries. We acknowledge a potential selection bias. Only hospitalized APE patients were considered for inclusion in this study; therefore, from the starting point, they had a higher overall risk of mortality, with most participants having PESI scores corresponding to class III or higher. Moreover, only patients with available ABG analysis were included, meaning that some APE patients that were admitted to the hospital, but did not fit into the ER protocol for ABG on admission, were not included in the study.