3.10 Sepsis: Critical Closing Pressure & Tissue Perfusion Pressure

Sepsis remains one of the most challenging syndromes in global critical care medicine. Its pathophysiological core lies in an uncontrolled inflammatory response that leads to vascular dysfunction, microcirculatory failure, and tissue hypoxia. Current international guidelines distill nearly four decades of clinical practice into a seemingly “simple” resuscitation target: maintaining a mean arterial pressure (MAP) ≥65 mmHg. This threshold is established based on extensive observational studies demonstrating that a MAP below 60–65 mmHg is closely associated with markers of organ hypoperfusion (e.g., elevated lactate, oliguria) and increased mortality [1]. However, clinicians frequently encounter a practical paradox: a substantial proportion of patients maintain a MAP within or even above the target range, yet continue to exhibit biochemical and clinical evidence of persistent tissue hypoxia. This phenomenon is particularly prominent in the ICU, highlighting a fundamental blind spot in traditional monitoring: MAP only reflects systemic macrovascular pressure but fails to reveal the actual pressure gradient driving capillary bed perfusion or its dynamic stability.

In 2025, a retrospective multicenter cohort study from China evaluated the prognostic value of critical closing pressure (Pcc) and tissue perfusion pressure (TPP) for mortality and acute kidney injury (AKI) in sepsis patients. It explored risk stratification metrics beyond MAP, providing a foundation for subsequent precision resuscitation strategies [2].

I. Core Content of the Study

(A) Study Background

A novel target for hemodynamic management in sepsis is TPP rather than MAP itself. TPP refers to the pressure gradient between MAP and the closing pressure of the downstream vascular system. The Starling resistor model provides a classic explanation: arteriolar smooth muscle tone establishes the Pcc of collapsible vessels. When arterial pressure falls below Pcc, the vessels collapse completely, and blood flow ceases. Therefore, effective TPP = MAP – Pcc, and its magnitude directly determines capillary blood flow [3]. In sepsis, inflammatory mediators (e.g., tumor necrosis factor-alpha [TNF-α] and interleukin-6 [IL-6]) excessively activate ATP-sensitive potassium channels in vascular smooth muscle, leading to a rapid loss of vascular tone and a significant decrease in Pcc. This alteration produces a dual detriment: (1) Although Pcc decreases, the drop in MAP is even more pronounced, resulting in a paradoxical decrease in TPP rather than an increase, thereby reducing the driving force for tissue perfusion; (2) The collapse of the vascular waterfall effect allows downstream venous pressure fluctuations to be directly transmitted to the arterial end, stripping the microcirculation of its pressure buffer. Consequently, blood flow exhibits “fragile” characteristics, becoming highly sensitive to minor drops in blood pressure.

Furthermore, this study enrolled adult sepsis patients (aged ≥18 years) admitted to the ICUs of 18 tertiary hospitals in China between 2013 and 2022. After an initial screening of 14,604 ICU admission records and applying strict exclusion criteria (ICU stay <24 hours, non-first admission, lack of continuous hemodynamic data, etc.), a total of 6,769 patients were included in the primary analysis cohort. The cohort exhibited typical sepsis demographic characteristics: a mean age of 61 years, 61.0% male, and a mean SOFA score of 7 (indicating moderate organ dysfunction). The primary sites of infection were pulmonary (37.1%) and abdominal (14.3%). Fifty-two point five percent (52.5%) of patients required vasoactive drug support. The overall ICU mortality rate was 26.4%, hospital mortality was 34.6%, and the incidence of AKI was 19.6% at 48 hours, rising to 25.9% within 7 days. These outcomes are consistent with international multicenter data, lending generalizability to the findings.

(B) Study Results

The primary results of this study revealed the distribution characteristics and interrelationships of a series of hemodynamic pressure parameters, representing the first time in real-world sepsis cohorts that the distribution profiles of Pcc and TPP were characterized: mean Pcc was 45 ± 12 mmHg, and mean TPP was 40 ± 12 mmHg. As Pcc quartile groups increased (<30, 30–45, 45–60, ≥60 mmHg), TPP showed a significant decreasing trend (P<0.001), visually confirming their inverse relationship under fixed MAP conditions. Furthermore, mean systemic filling pressure (Pmsf) remained stable across groups (12–14 mmHg), while the vascular waterfall gradient (VW = Pcc – Pmsf) increased significantly with rising Pcc, confirming that Pcc is the dominant variable regulating the waterfall effect in sepsis.

(C) Heterogeneity in Prognostic Stratification

The four-group prognostic model constructed based on optimized thresholds from ROC curves (Pcc 42 mmHg, TPP 38 mmHg) reveals distinct risk gradient phenotypes.

Low TPP–Low Pcc group (dual-low phenotype): ICU mortality was 35.1%, in-hospital mortality 44.4%, and 48-hour AKI incidence 33.1%. Patients in this group had a mean MAP of only 78 mmHg, but the core issue was a comprehensive loss of vascular tone, failing to maintain effective driving pressure and losing the pressure buffer of the waterfall effect.

High TPP–High Pcc group (dual-high phenotype): ICU mortality was 20.1%, in-hospital mortality 27.9%, and 48-hour AKI incidence 15.4%. This group had a mean MAP of 95 mmHg, with vascular tone and cardiac function in optimal coupling.

The two intermediate groups (Low TPP–High Pcc vs. High TPP–Low Pcc) both had mortality rates of 27%–28% with no statistical difference, suggesting that optimizing a single parameter is insufficient to improve prognosis and that dual-target synergy is required. After adjusting for SOFA score, APACHE II score, comorbidities, and MAP levels, the dual-low phenotype still increased the risk of death by 1.42–2.15 times (OR = 1.424–2.154, P < 0.001), confirming its prognostic value independent of disease severity. Pcc and TPP are independent prognostic indicators in sepsis patients; when used in conjunction with MAP, they improve risk stratification. Optimal organ perfusion and prognosis are achieved only when both exceed their thresholds.

In the subgroup maintaining MAP ≥80 mmHg, simply increasing Pcc (leading to decreased TPP) or simply increasing TPP (requiring higher MAP) did not reduce mortality; only the optimized combination of both (High TPP–High Pcc) yielded significant benefit.

(D) Dual Stratification Analysis to Unmask MAP Confounding Effects

To avoid confounding by MAP, the study employed an innovative matching-ranking stratification. In the MAP ≥80 mmHg subgroup, one variable (e.g., MAP) was fixed while the other (Pcc or TPP) was varied to observe prognostic changes. This method, previously used in driving pressure research for ARDS [4], efficiently isolates collinear variable effects and reveals independent associations.

II. Research Methodology and Academic Value

(A) Research Methodology

The study employed a multicenter retrospective cohort design. Building on internal validation (SEPSIS-EDT and PUMCH dual databases), the MIMIC-IV database was introduced for external validation. Inclusion criteria were precise and representative, comprising adults (≥18 years), first-time ICU admission, and confirmed sepsis (using Sepsis-3 criteria [5]). Exclusion criteria effectively eliminated confounders, including ICU stay <24 hours (to avoid insufficient data), non-first admission (to avoid repeated measurement bias), and those who actively refused treatment (to avoid confounding causes of death). The final sample size of 6,769 provided sufficient statistical power (>90%) to detect moderate effect sizes (OR ~1.5).

(B) Academic Value

Principles and validation of Pcc estimation methods, and the physiological basis of the mathematical model:Pcc estimation is based on the pressure-flow relationship in circulatory theory. In the arterial vascular bed, MAP = systemic vascular resistance × cardiac output + Pcc. This study used pulse pressure (PP) multiplied by heart rate (HR) as a non-invasive surrogate for cardiac output, based on the Windkessel model, where the aorta acts as an elastic reservoir and stroke volume is proportional to pulse pressure; thus, HR × PP approximates cardiac output. Based on this, researchers decomposed MAP into two key parameters via linear regression: (1) slope (k), corresponding to the coefficient of PP × HR, reflecting vascular resistance; and (2) intercept (b), which is Pcc. Within a 3-hour stable window, this simplified model is physiologically sound. Valid Pcc estimates require k > 0 (positive vascular resistance), b > 0 (Pcc as a physiological positive value), and a goodness-of-fit ≥ 0.5, ensuring reliability. Additionally, each patient yielded an average of 10 ± 4 valid measurements within 24 hours, a density that significantly reduces random error.

Design for accuracy validation: The study prospectively enrolled 17 patients with invasive arterial monitoring, comparing the 3-hour interval regression method (Method 1) with the 5-minute high-frequency data time-segmented regression method (Method 2). The Pcc estimates from both methods were highly consistent (r = 0.84, mean difference only 1 mmHg), confirming that time intervals have minimal impact on the mean and validating the robustness of the retrospective approach.

Application of Statistical Methods:

(1) Threshold determination method: Combining ROC curves with the Youden index, the optimal cutoff points for Pcc and TPP in the training set were determined as 42 mmHg and 38 mmHg, respectively. These thresholds shifted slightly in the validation set (Pcc to 38 mmHg, TPP to 33 mmHg), but the risk ranking of samples remained stable, confirming the robustness of the core conclusions.

(2) Dual stratification analysis: To avoid confounding by MAP, the study employed an innovative matching-ranking stratification. In the MAP ≥80 mmHg subgroup, one variable (e.g., MAP) was fixed while the other (Pcc or TPP) was varied to observe prognostic changes. This method, previously used in driving pressure research for ARDS, efficiently isolates collinear variable effects and reveals independent associations.

(3) Application of competing risk models: When evaluating AKI, death as a competing event may lead to underestimation of AKI incidence. The study used the Fine-Gray model for adjustment, and results showed the main conclusions remained unchanged.

(4) Sensitivity analysis: To ensure the integrity of the evidence chain and robustness of conclusions, a multi-level, multi-dimensional sensitivity analysis system was designed, including: (1) complete-case analysis; (2) using the Liljestrand-Zander formula instead of PP × HR to estimate cardiac output; and (3) regression across different time windows (1/2/4 hours). All results showed consistent effect directions, stable statistical significance, and unchanged risk rankings, ruling out model dependency.

III. Research Limitations and Future Scientific Significance

Pcc/TPP estimation only requires monitoring invasive arterial blood pressure waveforms and heart rate, which modern monitors are fully capable of computing. Limitations of this study include data bias from its retrospective design, lack of individualized optimal thresholds, absence of organ-specific indicators, exclusion of pre-sepsis comorbidities, and limitations in cardiac output estimation. Therefore, future research should focus on threshold generalizability: ideal Pcc/TPP values may differ in elderly patients, those with chronic kidney disease, or heart failure. Since the current thresholds derive from a mixed population, future studies should refine analyses by subgroups. Additionally, attention should be given to dynamic temporal studies, interventional causal research, and RCTs to validate the efficacy of targeted interventions.

Blood pressure management is transitioning from a traditional MAP-dependent model to an era of precision hemodynamics. In this process, microcirculatory perfusion assessment is gradually becoming the core basis for clinical decisions. Although bridging the gap from evidence to practice will take time, the blueprint for precise resuscitation is already clear: future ICU monitors will no longer display just a single MAP number, but a dynamic MAP–Pcc–TPP triangular balance chart, guiding clinicians toward true tissue-perfusion-oriented therapy. For every intensivist facing the challenges of sepsis, the significance of this study lies not only in providing a new risk stratification tool but also in awakening profound reflection on the essence of microcirculatory tissue perfusion—blood pressure is the manifestation, while perfusion is the essence. Amidst the current dilemmas in septic shock treatment, this innovation returning to physiological fundamentals may well be the key to unlocking precision therapy.

(Du Wei, Peking Union Medical College Hospital)

References

  1. Singer M, Deutschman CS, Seymour CW, et al. The third international consensus definitions for sepsis and septic shock (Sepsis-3) [J]. JAMA, 2016, 315(8): 801-810.

  2. Wang JY, Diao ST, Zhu TY, et al. Critical closing and tissue perfusion pressures in sepsis—implications for risk stratification: a retrospective cohort study [J]. Anesthesiology, 2026, 144(4): 886-897.

  3. Castro R, Kattan E, Retamal J, et al. Venous congestion from a vascular waterfall perspective: reframing congestion as a dynamic Starling resistor phenomenon [J]. Intensive Care Med Exp, 2025, 13(1): 119.

  4. Sanchez EC, Taha A, Tolba Y, et al. Assessment of tissue perfusion pressure in patients with septic shock: beyond mean arterial pressure [J]. Crit Care Med, 2025, 53(11): e2305-e2317.

  5. Prescott HC, Antonelli M, Alhazzani W, et al. Surviving Sepsis Campaign: international guidelines for management of sepsis and septic shock 2026 [J]. Intensive Care Med, 2026. doi:10.1097/CCM.0000000000007075. Online ahead of print.