Abstract
Background:
Although Sepsis-3.0 defines septic shock as hypotension with serum lactate levels >2.0 mmol/L, this criterion may miss low-lactate shock: a clinically significant phenotype characterized by hypotension but without elevated lactate levels. The epidemiological characteristics and prognostic significance of low-lactate shock remain unclear, highlighting a critical gap in current shock management models.
Methods:
We conducted a retrospective cohort study of 3,134 patients with shock admitted to a tertiary care medical institution from January 2015 to March 2022. We used propensity score matching (1:2 ratio) to control for confounding factors, aiming to determine the prevalence of low-lactate shock (lactate ≤ 2.0 mmol/L), identify risk factors through multivariable logistic regression, and validate the predictive model [Nutritional Risk Screening-2002-Acute Physiology and Chronic Health Evaluation-Triglycerides-Total Bilirubin (NRS-APACHE II-TG-TBIL)] using receiver operating characteristic analysis.
Results:
The 28-day mortality rate was slightly lower in the low-lactate shock group compared with the high-lactate group [25.4% (94/369) vs. 35.8% (990/2,765), respectively]. The age, Nutritional Risk Screening-2002 (NRS-2002) score, and venous thromboembolism risk score were significantly lower in the low-lactate shock group than in the high-lactate shock group (P < 0.05). No statistically significant differences were observed in gender distribution (P = 0.092). Multivariable analysis identified four independent predictors of low-lactate shock: NRS-2002 [odds ratio (OR) = 0.570, P < 0.001], Acute Physiology and Chronic Health Evaluation II (APACHE II) (OR = 0.869, P < 0.001), triglycerides (OR = 0.772, P = 0.035), and total bilirubin (OR = 0.993, P = 0.002). The composite NRS-APACHE II-TG-TBIL model showed excellent discrimination (area under curve = 0.800, P < 0.001) with balanced sensitivity (72.6%) and specificity (73.5%).
Conclusions:
Low-lactate shock carries substantial mortality risk (25.4%). The validated NRS-APACHE II-TG-TBIL model (area under curve = 0.800) provides an effective tool for early detection, addressing critical diagnostic gaps in shock management.
Keywords: Lactate, mortality, Nutritional Risk Screening-2002, prognostic risk model, shock
INTRODUCTION
Shock represents a life-threatening clinical syndrome characterized by acute circulatory failure and metabolic dysfunction, resulting in inadequate cellular oxygen utilization. According to the 2014 European Society of Intensive Care Medicine consensus, shock is defined as acute circulatory failure accompanied by impaired tissue oxygenation, irrespective of blood pressure levels (1). This condition carries substantial global disease burden, with epidemiological data revealing approximately 48.9 million cases of septic shock worldwide in 2017, accounting for 11 million sepsis-related deaths, while hemorrhagic shock contributes to an additional 1.9 million annual fatalities (2,3). The Sepsis-3.0 criteria (2016) operationalize septic shock as infection-induced circulatory dysfunction requiring vasopressors to maintain mean arterial pressure ≥65 mmHg despite adequate fluid resuscitation, coupled with serum lactate levels >2 mmol/L (4). While these criteria have gained widespread adoption, ongoing debate persists regarding the diagnostic sensitivity of this lactate threshold, particularly in cases of low-lactate shock.
The pathophysiological hallmark of shock manifests as microcirculatory failure, where blood pressure and lactate serve as key but imperfect surrogate markers (5). Physiologically, arterial lactate concentrations range between 0.5 and 1.5 mmol/L in healthy individuals, while critically ill patients typically maintain levels ≤2.0 mmol/L (6,7). Notably, 15% - 20% of shock patients present with preserved lactate levels (≤2.0 mmol/L) despite demonstrable circulatory dysfunction and comparable mortality risks (8). This clinical paradox may arise through multiple mechanisms: 1) macrocirculatory dysfunction with microcirculatory shunting, 2) enhanced mitochondrial oxidative phosphorylation capacity increasing lactate clearance (9), 3) increased compensatory hepatic metabolism (10), or 4) distinct metabolic profiles in specific shock subtypes (e.g., neurogenic shock) (11).
Emerging evidence positions lactate as a more sensitive shock biomarker than hemodynamic parameters (12), with levels >2.0 mmol/L signaling transition from compensated to decompensated shock (1,13,14). Lactate levels >4.0 mmol/L are an independent risk factor for increased mortality (14). However, contemporary studies challenge the diagnostic specificity of this cutoff, demonstrating that low-lactate shock patients exhibit comparable mortality risks to high-lactate shock cases (15). These findings emphasize that although the Sepsis-3.0 lactate criteria have prognostic value, they should not be used as the absolute criterion for diagnosing shock. While following international guidelines, it is essential to combine them with a comprehensive clinical assessment, including tissue perfusion indicators (such as capillary refill time and skin mottling) and organ function evaluation, to avoid overlooking patients with low-lactate shock.
This investigation pursues three principal objectives: to quantify the incidence and mortality burden of low-lactate shock, to identify associated clinical risk factors through multivariable analysis, and to develop a novel predictive model integrating nutritional status [Nutritional Risk Screening-2002 (NRS-2002)], disease severity [Acute Physiology and Chronic Health Evaluation II (APACHE II)], and metabolic parameters [triglycerides (TG), bilirubin]. Specifically, we will explore differences in clinical characteristics and outcomes between patients with low-lactate shock and those with high-lactate shock, providing evidence to optimize current diagnostic criteria, thereby enabling timely intervention and accurate management of shock patients.
PATIENTS AND METHODS
Source of data
We conducted a retrospective cohort study utilizing clinical data from shock patients admitted to the Second Affiliated Hospital of Zhejiang University School of Medicine, a large tertiary medical institution in Hangzhou, China. The study period spanned from January 2015 to March 2022. All patient data were extracted from the hospital’s electronic medical record system (EMRS). The study protocol was reviewed and approved by the Institutional Ethics Committee of the Second Affiliated Hospital of Zhejiang University School of Medicine (approval no: 2024-0247). Due to the retrospective nature of the investigation and use of anonymized clinical data, the requirement for written informed consent was waived by the ethics committee.
Patient selection criteria
This study enrolled adult patients admitted between January 1, 2015, and March 31, 2022, for the following conditions: primary admission diagnosis of shock and serum lactate measurements available within 7 days before or after admission. For patients with multiple lactate measurements during this period, the highest recorded value was selected for analysis (see the detailed screening process in Supplementary Materials, https://links.lww.com/SHK/C613). Patients were excluded if they met any of the following criteria: age < 18 years, duplicate medical records, incomplete clinical data or missing key variables, unclear type of shock, shock had resolved at admission (hemodynamically stable without intervention). The complete patient screening and selection process is shown in Figure 1.
Fig. 1.
Flowchart of patient screening and selection process.
Data collection
Comprehensive patient data were extracted from the EMRS, encompassing: 1) demographic characteristics including age, sex, mode of admission, admission diagnosis, hospital department, length of hospital stay, and hospital outcome; 2) shock classification categorized as septic, hypovolemic, cardiogenic, neurogenic, anaphylactic, or traumatic; 3) standardized hospital assessment scores including venous thromboembolism (VTE), NRS-2002, and APACHE II; and 4) laboratory parameters comprising body mass index (kg/m²), hemoglobin (g/dL), platelet count (×10⁹/L), white blood cell count (×10⁹/L), albumin (g/L), creatinine (μmol/L), C-reactive protein (mg/L), total cholesterol (mmol/L), TG (mmol/L), and total bilirubin (TBIL, μmol/L). For patients with multiple measurement results, use the initial laboratory results to ensure a consistent baseline evaluation.
Statistical analysis
Statistical analyses were performed using appropriate parametric and nonparametric methods: continuous variables were expressed as mean ± standard deviation or median (interquartile range) and compared using Student’s t test or Wilcoxon rank-sum test, respectively; categorical variables were presented as frequencies (percentages) and analyzed using Pearson’s chi-square test without continuity correction. Survival analysis was performed using Kaplan–Meier methodology with between-group differences assessed by log-rank testing. Predictive modeling involved both univariate and multivariable logistic regression analyses, with results reported as odds ratios (ORs) with corresponding 95% confidence intervals (CIs), while model discrimination was evaluated through receiver operating characteristic curve analysis. To address potential confounding, propensity score matching (PSM) was implemented using Stata version 17.0 (StataCorp LLC, College Station, Texas). All statistical procedures were executed using SPSS Statistics version 25.0 (IBM Corporation, Armonk, New York), and a P value of <0.05 was considered statistically significant.
PSM analysis
To address potential confounding when comparing nutritional status between low-lactate [lactate (Lac) ≤ 2.0 mmol/L] and high-lactate (Lac > 2.0 mmol/L) shock patients, we conducted a PSM analysis, with propensity scores derived from multivariable logistic regression incorporating the primary exposure variable NRS-2002 along with relevant confounders including age, sex, shock type (categorized as septic, hypovolemic, cardiogenic, neurogenic, anaphylactic, or traumatic), and VTE risk score. Covariates were selected based on their established associations with both lactate metabolism and clinical outcomes in critical illness. Matching was performed using nearest-neighbor methodology with a caliper width of 0.2 standard deviations of the logit propensity score to ensure matching quality, while adopting a 1:2 matching ratio (low-lactate (vs.)high-lactate) to optimize statistical power. Postmatching balance was systematically evaluated using the standardized mean difference (SMD) and verified using Love plots and comparisons of differences before and after matching. An SMD value of less than 10% indicates good matching balance.
RESULTS
Distribution characteristics
The study cohort comprised 3,134 eligible shock patients and excluded 780 cases (Fig. 1). Stratification by lactate levels revealed 369 patients (11.8%) with normolactatemia (Lac ≤ 2.0 mmol/L), 1,003 (32.0%) with moderate hyperlactatemia (Lac: 2.1–4.0 mmol/L), and 1,762 (56.2%) with severe hyperlactatemia (Lac: 4.1–50.0 mmol/L). Shock subtype analysis demonstrated significant variation in lactate distribution patterns: anaphylactic shock (18.2%, 6/33) and hypovolemic shock (16.0%, 97/606) have a high low-lactate shock incidence, while cardiogenic shock exhibited the lowest prevalence of normolactatemia(10.6%, 44/416) coupled with the highest rate of severe hyperlactatemia (70.4%, 293/416). Septic shock, representing the largest subgroup, demonstrated intermediate proportions across all lactate categories (Table S1, https://links.lww.com/SHK/C613).
Mortality outcomes
Kaplan–Meier survival analysis revealed significant differences in 28-day mortality stratified by lactate levels (Fig. 2, A and B), with mortality rates demonstrating a graded increase corresponding to lactate elevation: 25.4% in the low-lactate group (Lac ≤ 2.0 mmol/L), 31.7% in moderate hyperlactatemia (Lac: 2.1–4.0 mmol/L), and 38.1% in severe hyperlactatemia (Lac: 4.1–50.0 mmol/L). Notably, the low-lactate group (Lac ≤ 2.0 mmol/L) accounted for approximately one-quarter of total mortality events despite representing only 11.8% of the study population, with their absolute mortality rate reaching 25.4%, underscoring the substantial mortality risk even in the absence of traditional lactate-defined shock criteria.
Fig. 2.
(A),28-day survival curves of patients in two groups; (B) 28-day survival curves of patients in three groups.
Baseline characteristics of patients
Initial comparative analysis of high-lactate subgroups (Lac: 2.1–4.0 vs. 4.1–50.0 mmol/L) revealed no statistically significant differences in baseline characteristics (Table S2, https://links.lww.com/SHK/C613), though mortality rates were markedly elevated in the severe hyperlactatemia cohort (Lac > 4.0 mmol/L). Given this clinical homogeneity, we consolidated high-lactate groups (Lac > 2.0 mmol/L) for subsequent analyses to minimize confounding. Table 1 shows that a comparative assessment between low-lactate (Lac ≤ 2.0 mmol/L) and high-lactate (Lac > 2.0 mmol/L) shock patients demonstrated significant intergroup differences: the low-lactate cohort exhibited a younger mean age (60.3 ± 17.8 vs. 62.9 ± 16.6 years, P = 0.004), lower VTE scores [5 (4–7) vs. 6 (4–8), P < 0.001], and reduced NRS-2002 risk [2 (1–3) vs. 3 (1–4), P < 0.001]. While gender distribution varied across lactate groups, these differences lacked statistical significance (P = 0.738). Multivariable logistic regression of the entire cohort (N = 3,134) identified VTE score (OR = 0.924, 95% CI: 0.891–0.959, P < 0.001) and NRS-2002 score (OR = 0.894, 95% CI: 0.843–0.950, P < 0.001) as independent predictors of lactate stratification, whereas age failed to demonstrate a significant association (Table 2).
Table 1.
Baseline characteristics stratified by low-lactate shock and high-lactate shock
| Variable | Lac ≤ 2.0 (n = 369) | Lac > 2.0 (n = 2,765) | Statistic | P value |
|---|---|---|---|---|
| 28-day mortality, n (%) | 94 (25.4) | 990 (35.8) | χ2 = 6.09 | 0.014 |
| Age, mean ± SD | 60.3 ± 17.8 | 62.9 ± 16.6 | T = 2.86 | 0.004 |
| VTE, M (Q25–Q75) | 5 (4–7) | 6 (4–8) | Z = 5.52 | <0.001 |
| NRS-2002, M (Q25–Q75) | 2 (1–3) | 3 (1–4) | Z = 5.31 | <0.001 |
| Sex (male), n (%) | 243 (65.9) | 1845 (66.7) | χ2 = 0.11 | 0.738 |
| Shock type, n (%) | χ2 = 12.23 | 0.032 | ||
| Septic shock | 200 (54.2) | 1,542 (55.8) | ||
| Hypovolemic shock | 97 (26.3) | 509 (18.4) | ||
| Cardiogenic shock | 44 (11.9) | 372 (13.4) | ||
| Neurogenic shock | 17 (4.6) | 262 (9.5) | ||
| Anaphylactic shock | 6 (1.6) | 27 (1.0) | ||
| Traumatic shock | 5 (1.4) | 53 (1.9) |
Lac, lactate; M, median; NRS-2002, Nutritional Risk Screening 2002; Q25, quantile 25; Q75, quantile 75; SD, standard deviation; VTE, venous thromboembolism.
Table 2.
Regression analysis of baseline characteristics of low-lactate shock (N = 3,134)
| Variable | Univariable analysis | Multivariable analysis | ||||
|---|---|---|---|---|---|---|
| OR | 95% CI | P value | OR | 95% CI | P value | |
| Age | 0.991 | 0.985–0.997 | 0.004 | 0.996 | 0.989–1.003 | 0.229 |
| Sex | 0.962 | 0.765–1.209 | 0.738 | 1.012 | 0.803–1.227 | 0.917 |
| VTE | 0.903 | 0.871–0.936 | <0.001 | 0.924 | 0.891–0.959 | <0.001 |
| NRS-2002 | 0.859 | 0.811–0.909 | <0.001 | 0.894 | 0.843−0.950 | <0.001 |
CI, confidence interval; NRS-2002, Nutritional RiskScreening; OR, odds ratio; VTE, venous thromboembolism.
PSM results
Through PSM at a 1:2 ratio, we successfully matched 369 low-lactate (Lac ≤ 2.0 mmol/L) patients with 738 high-lactate (Lac > 2.0 mmol/L) patients. The Love plot (Fig. 3) demonstrated excellent covariate balance, with all SMDs < 10%. Compared with the prematching group (N = 3,134), the postmatching cohort (n = 1,107) showed no significant differences in baseline characteristics, including age, gender distribution, type of shock, and VTE score, confirming the effective elimination of confounding factors (Table 3).
Fig. 3.
The love.plot before and after PSM.
Table 3.
Comparison of baseline characteristics differences before and after PSM
| Variable | Before propensity score match | After propensity score match | ||||
|---|---|---|---|---|---|---|
| Lac ≤ 2.0 (n = 369) | Lac > 2.0 (n = 2,765) | P value | Lac ≤ 2.0 (n = 369) | Lac > 2.0 (n = 738) | P value | |
| Age, mean ± SD | 60.26 ± 17.77 | 62.91 ± 16.56 | 0.004 | 60.26 ± 17.77 | 60.20 ± 17.67 | 0.973 |
| Sex (male), n (%) | 243 (65.9) | 1,845 (66.7) | 0.738 | 243 (65.9) | 486 (65.9) | >0.999 |
| VTE, M (Q25–Q75) | 5 (4–7) | 6 (4–8) | <0.001 | 5 (4–7) | 5 (4–7) | 0.869 |
| Shock type, n (%) | <0.001 | 0.947 | ||||
| Septic shock | 200 (54.2) | 1,542 (55.8) | 200 (54.2) | 405 (54.9) | ||
| Hypovolemic shock | 97 (26.3) | 509 (18.4) | 97 (26.3) | 194 (26.3) | ||
| Cardiogenic shock | 44 (11.9) | 372 (13.5) | 44 (11.9) | 88 (11.9) | ||
| Neurogenic shock | 17 (4.6) | 262 (9.5) | 17 (4.6) | 36 (4.9) | ||
| Anaphylactic shock | 6 (1.6) | 27 (1.0) | 6 (1.6) | 7 (0.9) | ||
| Traumatic shock | 5 (1.4) | 53 (1.9) | 5 (1.4) | 8 (1.1) | ||
Lac, lactate; M, median; PSM, propensity score matching; Q25, quantile 25; Q75, quantile 75; SD, standard deviation; VTE, venous thromboembolism.
Construction and validation of the prediction model
Comparative analysis of matched cohorts (Table 4) identified significant variables that were subsequently incorporated into univariate and multivariable logistic regression analyses. Based on existing research findings, through backward stepwise elimination, while accounting for clinical relevance and minimizing multicollinearity, four key predictors were selected for the final model: NRS-2002 score (OR = 0.570, 95% CI: 0.473–0.687, P < 0.001), APACHE II score (OR = 0.869, 95% CI: 0.841–0.899, P < 0.001), TG (OR = 0.772, 95% CI: 0.607–0.981, P = 0.035), and TBIL (OR = 0.993, 95% CI: 0.988–0.997, P = 0.002). The results demonstrate that lower values of these parameters reflect better nutritional status (NRS-2002), less severe physiological derangement (APACHE II), reduced lipid metabolism (TG), and preserved hepatic function (TBIL), which were independently associated with an increased incidence of low-lactate shock (Table 5).
Table 4.
Comparison of clinical features after propensity score matching
| Variable | Lac ≤ 2.0 (n = 369) | Lac > 2.0 (n = 738) | Statistic | P value |
|---|---|---|---|---|
| NRS-2002 | 2 (1–3) | 3 (2–4) | Z = 14.398 | <0.001 |
| APACHE II | 16 (11–21) | 24 (18–29) | Z = 17.355 | <0.001 |
| BMI (kg/m²) | 21.42 ± 3.67 | 22.14 ± 3.68 | T = 3.815 | <0.001 |
| DM | 74 (20.1) | 178 (24.1) | χ2 = 4.769 | 0.128 |
| HB (g/dL) | 8.1 (6.3–10.0) | 6.6 (5.4–8.7) | Z = 8.536 | <0.001 |
| PLT (109/L) | 110.0 (70.0–161.5) | 66.0 (29.0–116.3) | Z = 11.021 | <0.001 |
| WBC (109/L) | 14.1 (10.3–19.5) | 17.7 (13.0–24.2) | Z = 8.461 | <0.001 |
| CRP (mg/L) | 130.0 (63.9–208.0) | 167.2 (79.1–239.0) | Z = 5.487 | <0.001 |
| CR (µmol/L) | 90.0 (65.0–150.5) | 131.0 (84.0–234.3) | Z = 8.139 | <0.001 |
| ALB (g/L) | 26.0 (23.1–29.3) | 24.4 (20.6–27.6) | Z = 7.535 | <0.001 |
| TBIL (µmol/L) | 22.0 (14.5–33.8) | 31.1 (19.8–62.8) | Z = 8.981 | <0.001 |
| TG (mmol/L) | 1.15 (0.81–1.52) | 1.25 (0.89–1.71) | Z = 3.035 | 0.002 |
Normal distribution variables: mean ± SD; non-normal distribution variables: M (Q25–Q75); categorical variables: n (%).
ALB, albumin; BMI, body mass index; CR, creatinine; CRP, C-reactive protein; DM, diabetes mellitus; HB, hemoglobin; Lac, lactate; M, median; NRS-2002, Nutritional Risk Screening 2002; PLT, platelets; Q25, quantile 25; Q75, quantile 75; SD, standard deviation; TBIL, total bilirubin; TG, triglycerides; VTE, venous thromboembolism; WBC, white blood cells.
Table 5.
Regression analysis of predictors for low-lactate shock after PSM (n = 1,107)
| Variable | Univariate analysis | Multivariate analysis | ||||
|---|---|---|---|---|---|---|
| OR | 95% CI | P value | OR | 95% CI | P value | |
| NRS-2002 | 0.449 | 0.389–0.518 | <0.001 | 0.570 | 0.473–0.687 | <0.001 |
| APACHE II | 0.841 | 0.817–0.866 | <0.001 | 0.869 | 0.841–0.899 | <0.001 |
| HB | 1.020 | 1.014–1.026 | <0.001 | |||
| PLT | 1.009 | 1.007–1.011 | <0.001 | |||
| WBC | 0.947 | 0.930–0.963 | <0.001 | |||
| CRP | 0.997 | 0.995–0.998 | <0.001 | |||
| ALB | 1.091 | 1.061–1.122 | <0.001 | |||
| CR | 0.998 | 0.997–0.999 | <0.001 | |||
| TBIL | 0.990 | 0.986–0.994 | <0.001 | 0.993 | 0.988–0.997 | 0.002 |
| TG | 0.804 | 0.671–0.962 | 0.017 | 0.772 | 0.607–0.981 | 0.035 |
ALB, albumin; APACHE II, Acute Physiology and Chronic Health Evaluation II; BMI, body mass index; CI, confidence interval; CR, creatinine; CRP, C-reactive protein; HB, hemoglobin; NRS-2002, Nutritional Risk Screening 2002; OR, odds ratio; PLT, platelets; PSM, propensity score matching; TBIL, total bilirubin; TG, triglycerides; WBC, white blood cells.
Receiver operating characteristic analysis demonstrated variable predictive capabilities for individual parameters: the NRS-2002 score showed moderate discrimination [area under curve (AUC) = 0.687, 95% CI: 0.653–0.721) with high sensitivity (0.817) but limited specificity (0.485) at a cutoff of ≤2.5, while the APACHE II score exhibited stronger performance (AUC = 0.763, 95% CI: 0.734–0.792) with balanced sensitivity (0.665) and specificity (0.726) at 20.5 points. Metabolic markers displayed weaker but significant predictive value: TG (AUC = 0.552, 95% CI: 0.513–0.592; cutoff: 1.53 mmol/L) and TBIL (AUC = 0.644, 95% CI: 0.610–0.677; cutoff: 30.5 μmol/L). The composite NRS-APACHE II-TG-TBIL model achieved superior diagnostic accuracy (AUC = 0.801, 95% CI: 0.771–0.831) with optimal sensitivity (0.710) and specificity (0.745), confirming its robust discriminative capacity for low-lactate shock identification (Table 6 and Fig. 4).
Table 6.
Predictive performance of different parameters in low-lactate shock (n = 1,107)
| Variable | AUC (95% CI) | Sensitivity | Specificity | Cutoff value |
|---|---|---|---|---|
| NRS-2002 | 0.687 (0.653–0.721) | 0.817 | 0.485 | 2.5 |
| APACHE II | 0.763 (0.734–0.792) | 0.665 | 0.726 | 20.5 |
| TG (mmol/L) | 0.552 (0.513–0.592) | 0.366 | 0.752 | 1.53 |
| TBIL (mmol/L) | 0.644 (0.610–0.677) | 0.518 | 0.715 | 30.5 |
| NRS + APACHE II + TG | 0.791 (0.763–0.818) | 0.702 | 0.714 | |
| NRS + APACHE II + TBIL | 0.792 (0.764–0.819) | 0.696 | 0.735 | |
| NRS + APACHE II + TG + TBIL | 0.801 (0.771–0.831) | 0.710 | 0.745 |
APACHE II, Acute Physiology and Chronic Health Evaluation II; AUC, area under curve; CI, confidence interval; NRS-2002, Nutritional Risk Screening 2002; TBIL, total bilirubin; TG, triglycerides.
Fig. 4.
ROC curve of different parameters in low-lactate shock.
DISCUSSION
The concept of “low-lactate shock” has emerged to describe critically ill patients who present with shock despite having normal blood lactate levels (15). Several pivotal studies have established this clinical entity. For instance, Lebaron and Leonard’s (16) 2017 case report described septic shock without hyperlactatemia, while an analysis of the Surviving Sepsis Campaign database involving 19,945 septic shock patients revealed that 20.1% (4,000/19,945) did not exhibit hypotension during shock and 32.4% (6,460/19,945) had lactate levels ≤2.0 mmol/L within 6 hours of shock onset. Surprisingly, 6.53% (1,302/19,945) presented with neither hypotension nor hyperlactatemia during shock, yet their hospital length of stay and mortality rates were comparable to other shock patients (17). Our analysis of 3,134 shock patients identified an 11.8% incidence of low-lactate shock with 25.4% 28-day mortality, consistent with previous reports. Further supporting the clinical significance of low-lactate shock, a retrospective study from a Seoul hospital demonstrated that shock patients with lactate levels ≤2.0 mmol/L had significantly higher 28-day mortality than stroke and cardiac patients, with mortality independently associated with higher APACHE II scores, elevated C-reactive protein, and chronic heart failure (18). Similarly, Nichol et al. (15) found that among shock patients with admission or time-weighted lactate levels of 0.75–2.0 mmol/L, higher lactate concentrations correlated with increased in-hospital mortality, identifying lactate as an independent risk factor. Moreover, compared with patients with lactate ≤1.4 mmol/L, those with 1.4–2.3 mmol/L exhibited more severe organ dysfunction, poorer response to fluid resuscitation and vasopressors, and worse prognosis (19). These findings collectively suggest that even modest lactate elevations within the “normal” range may indicate high-risk shock phenotypes, challenging the traditional reliance on hyperlactatemia as a diagnostic and prognostic marker in shock.
Traditionally viewed as a metabolic byproduct of anaerobic glycolysis, lactate is now recognized as a key metabolic mediator with multifaceted roles in inflammation modulation, histone lactylation, cellular signaling, and antitumor activity (20–23). Emerging evidence indicates that lactate post-translationally modifies proteins such as PKM2 and METTL3, thereby influencing energy metabolism and signal transduction under conditions of cellular stress and inflammation (24,25). Beyond its metabolic functions, lactate acts as a signaling molecule capable of mitochondrial entry, where it activates the electron transport chain, shifts energy production from glycolysis to oxidative phosphorylation, and serves as an alternative carbon source to preserve glucose (9). Given its dual role as a metabolic regulator and energy substrate, circulating lactate levels are closely linked to nutritional status, as quantified by the NRS-2002 score, which is a validated tool for assessing malnutrition in hospitalized populations. Elevated NRS-2002 scores are independently associated with adverse clinical outcomes, including prolonged hospitalization, increased in-hospital mortality, and higher incidence of nonventilator-associated pneumonia. Consistent with these findings, our study identifies high NRS-2002 scores as an independent predictor of elevated lactate levels in patients with shock, underscoring the interplay between metabolic dysregulation and nutritional status in critical illness.
While prior studies have predominantly examined the prognosis of low-lactate shock patients (26–29), investigations into its underlying etiology and predictive potential remain limited. This study advances the field by delineating the clinical characteristics of low-lactate shock and developing a predictive model for this condition. Our findings reveal significant associations between low-lactate shock and admission NRS-2002 scores, APACHE II scores, TG, TBIL, and lactate metabolism, while these relationships have not been previously established in the literature. Our observations provide novel pathophysiological insights, corroborating and extending earlier reports. For instance, Doshi et al. (8) demonstrated that septic shock patients with lactate <2.5 mmol/L exhibited significantly lower APACHE II scores compared with those with levels ≥2.5 mmol/L [21 (15–27) vs. 27 (20–35), P < 0.001], while Dugas et al. (30) observed that 45% (55/123) of septic shock patients presented with initial lactate levels ≤2.4 mmol/L, with the highest proportion in the 1.5–2.0 mmol/L range, influenced by factors such as liver disease, bacteremia, bilirubin, hepatic enzymes, leukocyte count, and International Normalized Ratio.
The lower NRS-2002 scores in low-lactate shock patients may signify preserved metabolic compensatory mechanisms. Well-nourished individuals (low NRS-2002) likely maintain robust hepatic lactate clearance via an intact Cori cycle, given the established link between hepatic dysfunction and hyperlactatemia (31,32). Moreover, their efficient mitochondrial oxidative phosphorylation reduces dependence on anaerobic glycolysis (33–35). Conversely, malnourished patients (high NRS-2002) experience accelerated muscle catabolism and glycogen depletion, predisposing them to anaerobic metabolism and elevated lactate (36). Adequate branched-chain amino acid availability in low NRS-2002 patients may further support alternative energy pathways (37–39). The inverse relationship between TG and low-lactate shock underscores a potential metabolic reprogramming: inhibited fat oxidation during severe stress, which is reflected by reduced TG levels, may arise from blunted catecholamine responses (40,41) or diminished lipoprotein lipase activity (42), diverting energy production toward ketogenesis or proteolysis and thereby attenuating lactate accumulation (43,44). Additionally, preserved hepatic gluconeogenesis in patients with normal/mildly elevated TBIL could sustain glucose homeostasis, further modulating lactate levels. These mechanisms align with contemporary perspectives on lactate as both a metabolic intermediate and a signaling molecule, collectively shaping its accumulation dynamics in shock states (9,20).
Our four-parameter predictive model (NRS-2002, APACHE II, TG, TBIL) offers a clinically actionable framework for risk stratification in patients with suspected shock and lactate ≤2.0 mmol/L. Designed for rapid deployment in emergency and intensive care unit settings, this model incorporates specific prognostic thresholds: NRS-2002 ≤ 3, APACHE II ≤ 21, TG < 1.53 mmol/L, and TBIL < 30.5 μmol/L, which can be seamlessly integrated into an EMRS to facilitate early intervention. Nutritional management should be tailored to individual risk profiles: patients with lower NRS-2002 scores may benefit from medium-chain TG supplementation to address impaired fat oxidation (45,46), whereas those with higher scores require early initiation of branched-chain amino acid-enriched enteral nutrition (38,47). Additionally, vigilant glucose monitoring is warranted in patients with mildly elevated bilirubin levels (17–34 μmol/L) to mitigate metabolic dysregulation. Although low-lactate shock is associated with lower mortality compared with high-lactate shock (25.4% vs. 35.8%), this population remains at significant risk for adverse outcomes, necessitating prioritized microcirculatory assessment and prolonged monitoring to prevent delayed organ failure.
The dynamic nature of lactate metabolism in shock necessitates special consideration of a 7-day assessment window. While traditional resuscitation protocols prioritize early lactate clearance, our extended observation period serves two critical purposes: capturing the complete lactate trajectory to prevent misclassification of transient hyperlactatemia and identifying a distinct subgroup of patients (16.1%, 632/3914) who maintain persistently low lactate levels (≤2.0 mmol/L) despite meeting shock criteria. Sensitivity analyses using a narrower 3-day pre- and postadmission window revealed an even higher prevalence (20.6%, 745/3610), confirming that the 7-day window enhances diagnostic specificity while preserving the robustness of our primary findings. Notably, more than 75% of included patients underwent initial lactate measurement within 1 hour of admission, adhering to current sepsis management guidelines (48). This methodological approach mitigates potential concerns regarding temporal obscuration of true lactate dynamics while uncovering an underrecognized high-risk population.
This study represents a significant advancement in the field through the development and validation of a high-performance predictive model (AUC = 0.800) that integrates multiple clinical parameters. The model’s performance surpassed previous studies that analyzed single parameters (8,30), and its composite structure reflects the multifactorial nature of lactate metabolism in shock. While these findings provide important clinical insights, several limitations must be acknowledged, including potential selection bias inherent to the single-center retrospective design, possible reduction in predictive accuracy due to lactate level categorization, and the need for external validation in diverse patient populations. Future investigations should focus on several key areas: development of dynamic, time-weighted lactate indicators to better capture metabolic trajectories; detailed analysis of lactate clearance kinetics; multicenter validation studies and mechanistic exploration of the nutritional metabolites with lactate relationship to enable more precise shock phenotype classification.
CONCLUSIONS
This study establishes low-lactate shock (11.8% prevalence, 25.4% mortality) as a clinically significant entity. The NRS-APACHEII-TG-TBIL model (AUC = 0.800) provides a robust tool for early identification, addressing critical gaps in shock diagnosis and enabling timely intervention. These findings challenge conventional lactate-centric diagnostic modes while offering actionable clinical thresholds for improved decision-making.
Supplementary Material
ABBREVIATIONS
- APACHE II—
- Acute Physiology and Chronic Health Evaluation II
- AUC—
- area under the curve
- CI—
- confidence interval
- EMRS—
- electronic medical record system
- NRS-2002—
- Nutritional Risk Screening-2002
- NRS-APACHE II-TG-TBIL—
- Nutritional Risk Screening-2002-Acute Physiology and Chronic Health Evaluation-Triglycerides-Total Bilirubin
- OR—
- odds ratio
- PSM—
- propensity score matching
- SMD—
- standardized mean difference
- TBIL—
- total bilirubin
- TG—
- triglycerides
- VTE—
- venous thromboembolism
Author contributions: Z.Z. designed the study, revised the manuscript, and gave final approval of the version to be published; X.X. collected and analyzed the data and coordinated the writing and preparation of the manuscript; F.H. collected and analyzed the data. All authors read and approved the submitted version.
The authors report no conflicts of interest.
This research was supported by the National Key Research and Development Program of China, Grant No. 2021YFC2501800 (to Z.Z.) and the National Natural Science Foundation of China, Grant No. 82272182, 82072202 (to Z.Z.).
Supplemental digital content is available for this article. Direct URL citation appears in the printed text and is provided in the HTML and PDF versions of this article on the journal’s Web site (www.shockjournal.com).
REFERENCES
- 1.Cecconi M, De Backer D, Antonelli M, et al. Consensus on circulatory shock and hemodynamic monitoring. Task force of the European Society of Intensive Care Medicine. Intensive Care Med. 2014;40(12):1795–1815. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Rudd KE, Johnson SC, Agesa KM, et al. Global, regional, and national sepsis incidence and mortality, 1990-2017: analysis for the Global Burden of Disease Study. Lancet. 2020;395(10219):200–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Cannon JW. Hemorrhagic shock. N Engl J Med. 2018;378(4):370–379. [DOI] [PubMed] [Google Scholar]
- 4.Singer M, Deutschman CS, Seymour CW, et al. The Third International Consensus definitions for sepsis and septic shock (sepsis-3). JAMA. 2016;315(8):801–810. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Blumlein D, Griffiths I. Shock: aetiology, pathophysiology and management. Br J Nurs. 2022;31(8):422–428. [DOI] [PubMed] [Google Scholar]
- 6.Vincent JL, De Backer D. Circulatory shock. N Engl J Med. 2013;369(18):1726–1734. [DOI] [PubMed] [Google Scholar]
- 7.De Backer D. Lactic acidosis. Minerva Anestesiol. 2003;69(4):281–284. [PubMed] [Google Scholar]
- 8.Doshi PB, Park AY, Banuelos RC, et al. The incidence and outcome differences in severe sepsis with and without lactic acidosis. J Emerg Trauma Shock. 2018;11(3):165–169. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Cai X, Ng CP, Jones O, et al. Lactate activates the mitochondrial electron transport chain independently of its metabolism. Mol Cell. 2023;83(21):3904–3920.e7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Zong Z, Ren J, Yang B, et al. Emerging roles of lysine lactyltransferases and lactylation. Nat Cell Biol. 2025;27(4):563–574. [DOI] [PubMed] [Google Scholar]
- 11.Kwon BK, Tetreault LA, Martin AR, et al. A clinical practice guideline for the management of patients with acute spinal cord injury: recommendations on hemodynamic management. Global Spine J. 2024;14(3_suppl):187S–211S. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kiyatkin ME, Bakker J. Lactate and microcirculation as suitable targets for hemodynamic optimization in resuscitation of circulatory shock. Curr Opin Crit Care. 2017;23(4):348–354. [DOI] [PubMed] [Google Scholar]
- 13.Wieruszewski PM, Khanna AK. Vasopressor choice and timing in vasodilatory shock. Crit Care. 2022;26(1):76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Chebl RB, Tamim H, Dagher GA, et al. Serum lactate as an independent predictor of in-hospital mortality in intensive care patients. J Intensive Care Med. 2020;35(11):1257–1264. [DOI] [PubMed] [Google Scholar]
- 15.Nichol AD, Egi M, Pettila V, et al. Relative hyperlactatemia and hospital mortality in critically ill patients: a retrospective multi-centre study. Crit Care. 2010;14(1):R25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Lebaron J, Leonard RJ. Septic shock without hyperlactatemia. Am J Emerg Med. 2017;35(6):934.e3–934.e4. [DOI] [PubMed] [Google Scholar]
- 17.Casserly B, Phillips GS, Schorr C, et al. Lactate measurements in sepsis-induced tissue hypoperfusion: results from the Surviving Sepsis Campaign database. Crit Care Med. 2015;43(3):567–573. [DOI] [PubMed] [Google Scholar]
- 18.Oh DH, Kim MH, Jeong WY, et al. Risk factors for mortality in patients with low lactate level and septic shock. J Microbiol Immunol Infect. 2019;52(3):418–425. [DOI] [PubMed] [Google Scholar]
- 19.Wacharasint P, Nakada TA, Boyd JH, et al. Normal-range blood lactate concentration in septic shock is prognostic and predictive. Shock. 2012;38(1):4–10. [DOI] [PubMed] [Google Scholar]
- 20.Lin Y, Bai M, Wang S, et al. Lactate is a key mediator that links obesity to insulin resistance via modulating cytokine production from adipose tissue. Diabetes. 2022;71(4):637–652. [DOI] [PubMed] [Google Scholar]
- 21.Feng T, Zhao X, Gu P, et al. Adipocyte-derived lactate is a signalling metabolite that potentiates adipose macrophage inflammation via targeting PHD2. Nat Commun. 2022;13(1):5208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Li X, Yang Y, Zhang B, et al. Lactate metabolism in human health and disease. Signal Transduct Target Ther. 2022;7(1):305. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Rabinowitz JD, Enerbäck S. Lactate: the ugly duckling of energy metabolism. Nat Metab. 2020;2(7):566–571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Wu D, Spencer CB, Ortoga L, et al. Histone lactylation-regulated METTL3 promotes ferroptosis via m6A-modification on ACSL4 in sepsis-associated lung injury. Redox Biol. 2024;74:103194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Wang J, Yang P, Yu T, et al. Lactylation of PKM2 suppresses inflammatory metabolic adaptation in pro-inflammatory macrophages. Int J Biol Sci. 2022;18(16):6210–6225. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Fatemeh G, Fotsing G, Marques-Vidal P, et al. Predictive value of multiple variable models including nutritional risk score (NRS 2002) on mortality and length of stay of patients with COVID-19 infections. The INCOVO study. Clin Nutr ESPEN. 2023;55:357–363. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Cha JK, Kim HS, Kim EJ, et al. Effect of early nutritional support on clinical outcomes of critically ill patients with sepsis and septic shock: a single-center retrospective study. Nutrients. 2022;14(11):2318. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Chen Z, Wu H, Jiang J, et al. Nutritional risk screening score as an independent predictor of nonventilator hospital-acquired pneumonia: a cohort study of 67,280 patients. BMC Infect Dis. 2021;21(1):313. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Viana MV, Pantet O, Bagnoud G, et al. Metabolic and nutritional characteristics of long-stay critically ill patients. J Clin Med. 2019;8(7):985. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Dugas AF, Mackenhauer J, Salciccioli JD, et al. Prevalence and characteristics of nonlactate and lactate expressors in septic shock. J Crit Care. 2012;27(4):344–350. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Yao S, Chai H, Tao T, et al. Role of lactate and lactate metabolism in liver diseases (review). Int J Mol Med. 2024;54(1):59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Bartoloni B, Mannelli M, Gamberi T, et al. The multiple roles of lactate in the skeletal muscle. Cells. 2024;13(14):1177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Nuyttens L, Heyerick M, Heremans G, et al. Unraveling mitochondrial pyruvate dysfunction to mitigate hyperlactatemia and lethality in sepsis. Cell Rep. 2025;44(8):116032. [DOI] [PubMed] [Google Scholar]
- 34.Cheng Y, Guo L. Lactate metabolism and lactylation in kidney diseases: insights into mechanisms and therapeutic opportunities. Ren Fail. 2025;47(1):2469746. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Lu CL, Qin L, Liu HC, et al. Tumor cells switch to mitochondrial oxidative phosphorylation under radiation via mTOR-mediated hexokinase II inhibition--a Warburg-reversing effect. PLoS One. 2015;10(3):e0121046. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Olson B, Marks DL, Grossberg AJ. Diverging metabolic programmes and behaviours during states of starvation, protein malnutrition, and cachexia. J Cachexia Sarcopenia Muscle. 2020;11(6):1429–1446. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Nanno Y, Toyama H, Terai S, et al. Preoperative oral branched-chain amino acid supplementation suppresses intraoperative and postoperative blood lactate levels in patients undergoing major hepatectomy. JPEN J Parenter Enteral Nutr. 2019;43(2):220–225. [DOI] [PubMed] [Google Scholar]
- 38.Matsumoto K, Koba T, Hamada K, et al. Branched-chain amino acid supplementation increases the lactate threshold during an incremental exercise test in trained individuals. J Nutr Sci Vitaminol (Tokyo). 2009;55(1):52–58. [DOI] [PubMed] [Google Scholar]
- 39.De Palo EF, Gatti R, Cappellin E, et al. Plasma lactate, GH and GH-binding protein levels in exercise following BCAA supplementation in athletes. Amino Acids. 2001;20(1):1–11. [DOI] [PubMed] [Google Scholar]
- 40.Camell CD, Sander J, Spadaro O, et al. Inflammasome-driven catecholamine catabolism in macrophages blunts lipolysis during ageing. Nature. 2017;550(7674):119–123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Jocken JW, Blaak EE. Catecholamine-induced lipolysis in adipose tissue and skeletal muscle in obesity. Physiol Behav. 2008;94(2):219–230. [DOI] [PubMed] [Google Scholar]
- 42.Robin AP, Askanazi J, Greenwood MR, et al. Lipoprotein lipase activity in surgical patients: influence of trauma and infection. Surgery. 1981;90(2):401–408. [PubMed] [Google Scholar]
- 43.Fernández-Verdejo R, Mey JT, Ravussin E. Effects of ketone bodies on energy expenditure, substrate utilization, and energy intake in humans. J Lipid Res. 2023;64(10):100442. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Hui S, Cowan AJ, Zeng X, et al. Quantitative fluxomics of circulating metabolites. Cell Metab. 2020;32(4):676–688.e4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Maruyama T, Matsui S, Kobayashi R, et al. Medium-chain triglyceride-specific appetite is regulated by the β-oxidation of medium-chain fatty acids in the liver. Am J Physiol Endocrinol Metab. 2024;326(5):E735–E746. [DOI] [PubMed] [Google Scholar]
- 46.St-Onge MP, Jones PJ. Greater rise in fat oxidation with medium-chain triglyceride consumption relative to long-chain triglyceride is associated with lower initial body weight and greater loss of subcutaneous adipose tissue. Int J Obes Relat Metab Disord. 2003;27(12):1565–1571. [DOI] [PubMed] [Google Scholar]
- 47.Green CR, Alaeddine LM, Wessendorf-Rodriguez KA, et al. Impaired branched-chain amino acid (BCAA) catabolism during adipocyte differentiation decreases glycolytic flux. J Biol Chem. 2024;300(12):108004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Levy MM, Evans LE, Rhodes A. The Surviving Sepsis Campaign Bundle: 2018 update. Crit Care Med. 2018;46(6):997–1000. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.