EVOLUTION OF BLOOD LACTATE AND 90-DAY MORTALITY IN SEPTIC SHOCK. A POST HOC ANALYSIS OF THE FINNAKI STUDY

Transcription

1 SHOCK, Vol. 47, No. 5, pp , 2017 EVOLUTION OF BLOOD LACTATE AND 90-DAY MORTALITY IN SEPTIC SHOCK. A POST HOC ANALYSIS OF THE FINNAKI STUDY Elina Varis, * Ville Pettilä, * Meri Poukkanen, * Stephan M. Jakob, Sari Karlsson, Anders Perner, jj Jukka Takala, and Erika Wilkman *, the FINNAKI Study Group * Division of Intensive Care Medicine, Department of Anesthesiology, Intensive Care and Pain Medicine, University of Helsinki and Helsinki University Hospital, Helsinki, Finland; Inselspital, Bern University Hospital and University of Bern, Bern, Switzerland; Lapland Central Hospital, Rovaniemi, Finland; Tampere University Hospital, Tampere, Finland; and jj Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark Received 28 Jun 2016; first review completed 14 Jul 2016; accepted in final form 4 Oct 2016 ABSTRACT Hyperlactatemia predicts mortality in patients with sepsis and septic shock, and its normalization is a potential treatment goal. We investigated the association of blood lactate and its changes over time with 90-day mortality in septic shock. We performed a post hoc analysis of 513 septic shock patients with admission blood lactate measurements in the prospective, observational, multicenter FINNAKI study. Repetitive lactate measurements were available in 496 patients for analyses of change in lactate values during intensive care unit stay. The 90-day mortality for all patients was 33.3%. Patients with admission lactate >2 mmol/l had higher 90-day mortality than those with admission lactate 2 mmol/l (43.4% vs. 22.6%, P <0.001). Patients with persistent hyperlactatemia (>2 mmol/l) at 72 h had higher 90-day mortality compared with those with a lactate value of 2.0 mmol/l (52.0% vs. 24.3%, P <0.001). Time-weighted mean lactate values were higher in non-survivors than in survivors, (median [IQR] 2.05 [ ] mmol/l vs [ ] mmol/l, P <0.001). Time to normalization of lactate was comparable for 90-day non-survivors and survivors (median [IQR] 17.0 [ ] vs [ ] h, P ¼ 0.67). In separate models, time-weighted mean lactate, lactate value at 72 h, and hyperlactatemia at 72 h were independently associated with 90-day mortality, but admission lactate and time to normalization of lactate were not. These findings may inform future clinical trials using combined surrogate endpoints for mortality in septic shock patients. KEYWORDS Hemodynamics, lactate, mortality, outcome, sepsis, septic shock, vasopressor Address reprint requests to Erika Wilkman, MD, PhD, Division of Intensive Care Medicine, Department of Anesthesiology, Intensive Care and Pain Medicine, Helsinki University Hospital, PB 340, HUS, Helsinki, Finland. erika.wilkman@hus.fi FINNAKI Study Group: The FINNAKI study group: Central Finland Central Hospital: Raili Laru-Sompa, Anni Pulkkinen, Minna Saarelainen, Mikko Reilama, Sinikka Tolmunen, Ulla Rantalainen, Marja Miettinen. East Savo Central Hospital: Markku Suvela, Katrine Pesola, Pekka Saastamoinen, Sirpa Kauppinen. Helsinki University Central Hospital: Ville Pettilä, Kirsi-Maija Kaukonen, Anna-Maija Korhonen, Sara Nisula, Suvi Vaara, Raili Suojaranta-Ylinen, Leena Mildh, Mikko Haapio, Laura Nurminen, Sari Sutinen, Leena Pettilä, Helinä Laitinen, Heidi Syrjä, Kirsi Henttonen, Elina Lappi, Hillevi Boman. Jorvi Central Hospital: Tero Varpula, Päivi Porkka, Mirka Sivula Mira Rahkonen, Anne Tsurkka, Taina Nieminen, Niina Prittinen. Kanta-Häme Central hospital: Ari Alaspää, Ville Salanto, Hanna Juntunen, Teija Sanisalo. Kuopio University Hospital: Ilkka Parviainen, Ari Uusaro, Esko Ruokonen, Stepani Bendel, Niina Rissanen, Maarit Lång, Sari Rahikainen, Saija Rissanen, Merja Ahonen, Elina Halonen, Eija Vaskelainen. Länsi Pohja Central Hospital: Jorma Heikkinen, Timo Lavander, Kirsi Heinonen, Anne-Mari Juopperi. Middle Ostrobothnia Central Hospital: Tadeusz Kaminski, Fiia Gäddnäs, Tuija Kuusela, Jane Roiko. Satakunta Hospital district: Vesa Lund, Päivi Tuominen, Pauliina Perkola, Riikka Tuominen, Marika Hietaranta, Satu Johansson. South Karelia Central Hospital: Seppo Hovilehto, Anne Kirsi, Pekka Tiainen, Tuija Myllärinen, Pirjo Leino, Anne Toropainen. Tampere University Hospital: Anne Kuitunen, Jyrki Tenhunen, Ilona Leppänen, Markus Levoranta, Sanna Hoppu, Jukka Sauranen, Atte Kukkurainen, Samuli Kortelainen, Simo Varila. Turku University Hospital: Outi Inkinen, Niina Koivuviita, Jutta Kotamäki, Anu Laine. Oulu University Hospital: Tero Ala-Kokko, Jouko Laurila, Sinikka Sälkiö. Vaasa Central Hospital: Simo-Pekka Koivisto, Raku Hautamäki, Maria Skinnar. The authors report no conflicts of interest. 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 ( DOI: /SHK Copyright ß 2016 by the Shock Society INTRODUCTION Sepsis affects millions of people annually (1, 2). Septic shock, the most severe form of sepsis, is one of the major causes of death among critically ill patients (3, 4). Septic shock is associated with 90-day mortality risk of 20% to 45% and its incidence appears to be increasing (5, 6). The hemodynamic management of septic shock aims at correcting and preventing tissue hypoperfusion to reduce the risk of multiple organ dysfunction and death (7). Elevated blood lactate concentration is considered a surrogate marker of hypoperfusion and severity of sepsis (8, 9). Despite its limitations (10) hyperlactatemia is an independent predictor of mortality (11 13) and, thus, it was included in the most recent septic shock definition (14). Targeting lactate concentration to guide therapy has been suggested to reduce mortality (8, 15, 16), but data on the time course of lactatemia in septic shock patients receiving vasopressors are limited. We performed a post hoc analysis of septic shock patients in the prospective, observational, multicenter FINNAKI study database to evaluate the association of blood lactate and its changes over time with 90-day mortality in patients with septic shock. PATIENTS AND METHODS The Ethics Committee of the Helsinki University Hospital approved the study. Study inclusion was possible with deferred consent, and we obtained a written, informed consent from each patient or patient s proxy. 574

2 SHOCK MAY 2017 LACTATE AND MORTALITY IN SEPSIS AND SEPTIC SHOCK 575 arterial lactate values within 4 h of diagnosis of shock in the ICU (admission lactate) and of these, those patients with repeat recordings of arterial lactate from admission throughout the stay in ICU. We used pairwise deletion for missing values. Arterial lactate from admission to 72 h and mortality We identified all patients with lactate values at admission >2 mmol/l and 3 mmol/l for further analyses. The cut-off values were chosen based on earlier studies in septic and critically ill patients and the sepsis-3 criteria (14, 16, 20). For assessment of change in lactate during different time intervals, we identified patients having an arterial blood gas sample taken within 2 h of the end of the chosen period (0 to 24 2 h, 0 to 48 2 h, and 0 to 72 2 h). To assess the change in lactate from admission value to 72 h, we used the admission value and the first measured value at 72 h. For control calculation, we imputed time to death into calculation of normalization of lactate. Calculation of lactate-time integral is described in ESM appendix 1, lww.com/shk/a491. Time to normalization of lactate In patients with repetitive measurements with at least two arterial lactate values >2 mmol/l, we calculated the time between the first and the last lactate measurement >2 mmol/l, with possible normalized values in between, and the aggregate number of recordings exceeding 2 mmol/l for each patient. In patients with admission lactate levels of >2 mmol/l and 3 mmol/l, we calculated the time to reach the first value 2 mmol/l. FIG. 1. Study flowchart a g : 90-day mortalities of the patient populations of the flowchart: a 208/609 (34.2%), b 171/513 (33.3%), c 160/496 (32.3%), d 115/265 (43.4%), e 84/183 (45.9%), f 56/248 (22.6%), g 31/82 (37.8%). Patients The FINNAKI study (17) was a prospective, observational, multicenter study on acute kidney injury conducted in 17 Finnish intensive care units (ICUs) between September 1, 2011 and February 1, 2012, and it included 2,901 patients. The inclusion criteria for the FINNAKI study have been described elsewhere (17). From the main study database, we identified all patients from 15 ICUs with adequate vasopressor data and who fulfilled the definition for septic shock within 24 h of ICU admission (18), had blood lactate measurement at the time of diagnosis of septic shock, received vasopressor therapy during the first 24 h, and those with admission lactate of >2 mmol/l and 3 mmol/l respectively. The flow chart with patient groups and respective 90-day mortalities is shown in Figure 1. We also identified post hoc those patients who fulfilled the new septic shock definition (Sepsis-3) (14). All patients were treated and monitored in the participating ICUs at the discretion of treating clinicians. Definitions In the original study the 2001 ACCP/SCCM Consensus Conference definitions for sepsis and septic shock were used (18). In the present analysis, we also used the most recent definition for sepsis and septic shock (14). We defined and staged acute kidney injury by the Kidney Disease: Improving Global Outcomes criteria. Data collection Routine data (demographics, the International Classification of Diseases [ICD-10] diagnosis, ICU scores, physiologic measures, and outcomes), data on vasopressor and inotrope use, and all arterial blood gas samples (including arterial lactate measurements) drawn during intensive care treatment were collected prospectively during the FINNAKI study to the Finnish Intensive Care consortium database maintained by Tieto Ltd, Helsinki, Finland. The blood-gas samples in each participating ICU were drawn in a standardized way (2 ml sample) from arterial catheters and analyzed using point-of-care blood gas analyzers. The timing of blood samplings was at the discretion of ICU staff. Data analyses We identified patients with septic shock within 24 h of admission to the ICU (19). The attending physician who filled the study case report form diagnosed septic shock. Patients who were hypotensive and received vasopressors immediately before admission to the ICU were also defined at the time of the study as having septic shock. We then identified all patients with available recording of Interaction between lactate levels and NE dose We identified all patients receiving vasopressors during the first 5 days in the ICU based on data from the FINNAKI study and their maximum doses during the first 24 h. We assessed the association of use of norepinephrine during the first 5 days based on the protocol of the FINNAKI study and dose of norepinephrine during the first 24 h in the ICU with admission lactate and with 90-day mortality. Statistical analyses We present data as absolute numbers (percentages) or medians with interquartile range (IQR). We used the Mann Whitney U test to compare groups of continuous data and the chi-square or Fisher exact test, when appropriate, to compare groups of categorical data. We analyzed the ability of arterial lactate values at admission and that of arterial lactate at 72 h to predict 90-day mortality using area under receiver operating characteristics (ROC) curve (AUCs) with 95% confidence intervals (CI). Finally, we performed a backward logistic regression analysis to evaluate the possible independent associations between lactate and 90-day mortality using two separate populations including: all septic shock patients (18), and septic shock patients according to the new definition (14). We included prognostic factors with P < 0.1 into the separate multivariable backward logistic regression analyses (due to interaction) using admission lactate, lactate value at 72 h, timeweighted mean lactate, hours to normalization of lactate, and lactate >2 mmol/ L at 72 h (yes/no) (due to their interaction) as covariates, and 90-day mortality as the outcome variable. Finally, we reperformed all analyses using enter models to confirm the robustness of our results. All statistical analyses were performed using IBM, SPSS statistics 21.0 and 22.0 (IBM, Armonk, NY) or NCSS 8 (East Kaysville, Utah) software. RESULTS We identified 609 patients fulfilling the 2001 consensus criteria of septic shock within 24 h of ICU admission, and 513 with admission lactate measurement (Table 1). The 513 patients had a total of 13,520 arterial blood lactate measurements taken. Repetitive lactate measurements were done in 496 patients (97% of 513). The population of 496 patients with repetitive lactate measurements in the ICU was comparable to the whole study population (Table 1). A median of 40 (26 59) lactate values per patient was available for analysis for a median of 139 (94 142) h. Patients were diagnosed with septic shock at median of 0.37 h (IQR ) from admission to the ICU.

3 576 SHOCK VOL. 47, No. 5 VARIS ET AL. TABLE 1. Characteristics of the study patients Data available All septic shock, repetitive lactate values (n ¼ 496) 90-day survivors (n ¼ 336) 90-day non-survivors (n ¼ 160) P Data available All septic shock, with admission lactate (n ¼ 513) 90-day survivors (n ¼ 342) 90-day non-survivors (n ¼ 171) P Age ( ) 62.5 ( ) 70.0 ( ) < ( ) 62.0 ( ) 70.0 ( ) <0.001 Gender (male) (%) (61.3) 209 (62.2) 95 (59.4) (61.0) 212 (62.0) 101 (59.1) 0.57 Comorbidities 496 Diabetes (26.2) 90 (26.8) 40 (25.0) (25.7) 91 (26.6) 41 (24.0) 0.59 Hypertension (52.2) 166 (49.4) 93 (58.1) (52.0) 169 (49.4) 98 (57.3) 0.24 Chronic kidney disease (6.0) 14 (4.2) 16 (10.0) (6.0) 14 (4.1) 17 (9.9) Chronic liver disease (6.7) 12 (3.6) 21 (13.1) < (6.4) 12 (3.5) 21 (12.3) Postoperative admission (30.0) 113 (33.6) 36 (22.5) (29.8) 115 (33.6) 38 (22.2) APACHE ( ) 22.0 ( ) 30.0 ( ) < ( ) 22.0 ( ) 30.0 ( ) <0.001 SAPS 24 h (Median, IQR) ( ) 41.0 ( ) 59.0 ( ) < ( ) 41.0 ( ) 59.0 ( ) <0.001 SOFA score day 1 SOFA cardiovascular subscore day ( ) 9.0 ( ) 11.0 ( ) < ( ) 9.0 ( ) 11.0 ( ) < ( ) 4.0 ( ) 4.0 ( ) < ( ) 4.0 ( ) 4.0 ( ) <0.001 AKI (%) (d 1 5) (62.7) 194 (57.7) 117 (73.1) (62.0) 195 (57.0) 123 (71.9) KDIGO highest stage (d ( ) 1.0 ( ) 1.5 ( ) < ( ) 1.0 ( ) 1.0 ( ) < ) RRT (%) (20.6) 55 (16.4) 47 (29.4) (20.7) 56 (16.4) 50 (29.2) Vasopressors in ICU during first 24 h 496 Norepinephrine (%) (97.8) 329 (96.4) 156 (97.5) (97.7) 335 (98.0) 167 (97.7) 0.55 Dopamine (%) (1.6) 4 (1.2) 4 (2.5) (1.8) 4 (1.2) 5 (2.9) 0.17 Epinephrine (%) (6.7) 12 (3.6) 21 (13.1) < (7.0) 12 (3.5) 24 (14.0) <0.001 Any vasopressor (%) (99.0) * 332 (98.8) 159 (98.8) (99.0) * 338 (98.8) 170 (99.4) 0.67 (from cardiovascular SOFA score day 1) ICU length of stay Hospital length of stay Time to death (days) ( ) 5.0 ( ) 4.0 ( ) ( ) 4.0 ( ) 4.0 ( ) ( ) 15.0 ( ) 8.0 ( ) < ( ) 15.0 ( ) 8.0 ( ) < ( ) ( ) 8.0 ( ) < ( ) ( ) 8.0 ( ) <0.001 * Five patients (1.0%) received vasopressors only before ICU admission. Eight patients died 91 to 162 days after ICU admission. AKI indicates acute kidney injury; APACHE, acute physiology and chronic health evaluation score; ICU, intensive care unit; IQR, inter quartile range; KDIGO, Kidney Disease: Improving Global Outcomes; RRT, renal replacement therapy; SAPS II, simplified acute physiology score.

4 SHOCK MAY 2017 LACTATE AND MORTALITY IN SEPSIS AND SEPTIC SHOCK 577 TABLE 2. Lactate values in patients with septic shock divided by timing, level of lactatemia, and 90-day mortality Data available All (n ¼ 513) 90-day survivors (n ¼ 342) 90-day non-survivors (n ¼ 171) P Lactate >2 at admission, n (%) (51.7) 150/265 (56.6) 115/265 (43.4) <0.001 Lactate 2 at admission, n (%) (48.3) 192/248 (77.4) 56/248 (22.6) <0.001 Lactate 3 at admission, n (%) (35.6%) 99 /183 (54.1) 84/183 (45.9) <0.001 Pre-ICU lactate ( ) 2.30 ( ) 3.45 ( ) <0.001 Number of lactate measurements ( ) 17.0 ( ) 17.0 ( ) 0.23 during 72 h (n) Number of lactate measurements ( ) 2.0 ( ) 5.0 ( ) <0.001 >2 mmol/l during 72 h (n) Percentage of lactate ( ) 12.2 ( ) 57.7 ( ) <0.001 measurements >2 mmol/l (%) Time to normalization of lactate ( ) n ¼ ( ) n ¼ ( ) >2 mmol/l Time to normalization of lactate ( ) n ¼ ( ) n ¼ ( ) mmol/l (median/iqr) Time-weighted mean lactate ( ) 1.29 ( ) 2.05 ( ) <0.001 (median/iqr) Lactate time-integral 12 h, ( ) ( ) ( ) <0.001 (median/iqr) Lactate time-integral 24 h, ( ) ( ) ( ) <0.001 (median/iqr) Lactate time-integral 48 h, ( ) ( ) ( ) <0.001 (median/iqr) Lactate time-integral total for all values, (median/iqr) ( ) ( ) ( ) <0.001 Five patients (1.0%) received vasopressors only before ICU admission. For 17 (of 513, 3.3%) patients, the lactate values of septic shock were available only before ICU admission. ICU indicates intensive care unit. Mortality During the first 72 h, 53 of 513 (10.3%) died in the ICU while 122 (23.8%) patients were discharged alive. The 30-day mortality for all 513 patients was 28.5% (146/513) and the 90-day mortality was 33.3% (171/513). The 90-day mortality for patients with repetitive lactate measurements was comparable to that of the whole population, 32.3% (160/496). The 90- day mortality rates for patients with admission lactate of >2 mmol/l and 3.0 mmol/l at admission were 43.4% (115/265) and 45.9% (84/183) (Table 2). The AUC for admission lactate to predict 90-day mortality in ROC analysis was 0.67 (95% CI ). Table 2 shows lactate values and evolution of lactate values during the ICU stay and 90-day mortality. A time-weighted mean lactate median showed best prediction for 90-day mortality, with a positive likelihood ratio of 1.79 (95% CI ). Predictive values of admission lactate values, normalization of lactate, lactate-time integral for the first 24 h, and time weighted mean lactate are shown in Table 3. Change in lactate over time The overall resolution of hyperlactatemia was slower in the non-survivors compared with survivors (Figs. 1 3 ESM Figure 1, ESM Tables 1 and 2, The AUC for lactate at 72 2 h for 90-day mortality was 0.64 (95% CI ). Time to normalization of lactate The time to first normalization of lactate from hyperlactatemia (>2 mmol/l and 3 mmol/l) did not differ between 90- day survivors and non-survivors (Table 3). The time between the first and the last lactate recording >2 mmol/l, with possible normalized values in between, was however associated with 90- day mortality, 24.0 ( ) h in non-survivors versus 20.4 ( ) h in survivors (P ¼ 0.019). FIG. 2. Median lactate values in 6-h intervals for 90-day survivors and 90-day non-survivors during 0 to 72 h. Medians and IQR are presented in ESM Table 1, IQR indicates interquartile range.

5 578 SHOCK VOL. 47, No. 5 VARIS ET AL. FIG. 3. A, Quartiles of change in lactate value from admission to 72 h and 90-day mortality (mean, 95% CI for mean) in patients with lactate measurements at 72 h presenting with lactate over 2 mmol/l at admission (n ¼ 149). B, Quartiles of change in lactate value from admission to 72 h and 90- day mortality in all patients with lactate measurements at 72 h (n ¼ 301). Interaction between lactate levels and NE dose The maximum dose of norepinephrine during the first 24 h in the ICU was significantly lower in 90-day survivors than in non-survivors (0.15 mg/kg/min [ ] vs mg/kg/min [ ], P < 0.001). Independent predictors of mortality Independent predictors of 90-day mortality in septic shock patients (using backward logistic regression) for all patients are shown in Table 4. In separate models time-weighted mean lactate, the lactate value at 72 2 h, and lactate >2 mmol/l at 72 2 h were independently associated with 90-day mortality, but admission lactate (all available values) or time to first normalization of lactate were not. These associations were confirmed in a sensitivity analysis including only those 262 patients with admission lactate >2 mmol/l who fulfilled the new septic shock criteria (14) (ESM Table 3, SHK/A493) and using enter models (data not shown). DISCUSSION In this post hoc sub-study of the observational multicenter FINNAKI study, we found that lactate values at admission were associated with 90-day mortality in septic patients with hypotension in univariate, but not in multivariate analyses. In addition, we found that both hyperlactatemia (>2 mmol/l at 72 h), time-weighted mean lactate during ICU stay, but not time to first normalization of lactatemia were associated with 90-day mortality in septic shock patients. Incidence of hyperlactatemia and mortality In the present study, 52% of the patients had an arterial lactate level of over 2 mmol/l and 36% of the patients had an arterial lactate of 3 mmol/l at diagnosis of septic shock. The mortality rates of the current study are comparable with the mortality rates reported by the task force for developing new sepsis criteria (14, 21). The mortality rate of 45.9% in the current study was comparable to the mortality rate of 39% in patients with hyperlactatemia of 3 mmol/l in a mixed ICU population from the Surviving Sepsis Guidelines database (21). Arterial lactate at admission and at 72 h and mortality We found that admission lactate and lactate at 72 h predicted 90-day mortality relatively poorly by ROC analyses (AUCs of 0.67 and 0.73, respectively) (22). Our findings are well in line with previous studies. A retrospective analysis of septic shock patients reported an AUC value of 0.63 for baseline lactate to 28-day mortality in a multicenter material and TABLE 3. Prediction of 90-day mortality by admission lactate values, normalization time for patients with admission lactate values >2 and 3 mmol/l respectively (paramedian split), lactate-time integral (paramedian split), and time-weighted lactate value (paramedian split) Admission lactate >2 mmol/l Admission lactate 3 mmol/l Admission lactate >2 mmol/l median time to normalization Admission lactate 3 mmol/l median time to normalization median lactate time-integral 24 h median time weighted mean lactate Number (%) 265/513 (51.7) 183/513 (35.6) 101/201 (50.2) 64/128 (50.0) 182/365 (49.5) 248/496 (50.0) 90-day mortality 115/265 (43.4) 84/183 (45.9) 29/101 (28.7) 16/64 (25.1) 66/182 (36.2) 114/248 (46.0) (%) Sensitivity (CI 95%) 0.67 ( ) 0.49 ( ) 0.49 ( ) 0.47 ( ) 0.62 ( ) 0.71 ( ) Specificity (CI 95%) 0.56 ( ) 0.71 ( ) 0.49 ( ) 0.49 ( ) 0.55 ( ) 0.60 ( ) Positive likelihood ratio (CI 95%) 1.53 ( ) 1.70 ( ) 0.97 ( ) 0.92 ( ) 1.39 ( ) 1.79 ( )

6 SHOCK MAY 2017 LACTATE AND MORTALITY IN SEPSIS AND SEPTIC SHOCK 579 TABLE 4. Multivariable backward regression regarding associations between blood lactate and 90-day mortality Model 5 Lactate over 2 mmol/l at 72 h Model 4 h to normalization of lactate Model 3 Time-weighted mean lactate Model 2 Lactate at 72 h Model 1 Admission lactate All patients Variable OR (CI 95%) P OR (CI 95%) P OR (CI 95%) P OR (CI 95%) P OR (CI 95%) P Chronic liver disease 3.45 ( ) ( ) ( ( ) ( ) AKI; highest KDIGO stage 0.86 ( ) ( ) ( ) ( ) ( ) 0.31 SAPS II; 24 h 1.06 ( ) < ( ) < ( ) < ( ) < ( ) <0.001 SOFA score; day ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) 0.88 Epinephrine used during first 24 h 1.04 ( ) ( ) ( ) ( ) ( ) 0.20 Norepinephrine; highest dose during first 24 h 2.56 ( ) Admission lactate 0.98 ( ) 0.56 Lactate value at 72 h 1.47 ( ) Time weighted mean lactate 1.48 ( ) <0.001 Normalization of lactate; hours 1.00 ( ) 0.49 Lactate value over 2 mmol/l at 72 h 0.66 in a single-center material (23). These results were later confirmed in patients with severe sepsis or septic shock admitted to the emergency department (20). In the study by Puskarich et al. (20), admission lactate was the best predictor of hospital mortality with an AUC of Comparable results were also found in a large Australian retrospective study of critically ill patients (24). Lactate change over time The results of the present study indicate that failure to normalize lactate at 72 h is associated with 90-day mortality. For patients with values >2 mmol/l at 72 h the 90-day mortality was over 2-fold. Persistent hyperlactatemia and slow normalization of lactate values have been shown to predict adverse outcome and mortality in patients with septic shock as well as in other critically ill patient groups, and lactate clearance, i.e., a certain decrease of lactate over a certain time has been associated with outcome benefit (25 27). Recent studies have also indicated that intermediate levels of lactatemia and relatively small changes in lactate values (9, 11), even within the normal range, may be linked to outcome (11, 23, 28). There is also some evidence that dynamic lactate measures describing the evolution of lactate over time may be more important than single, static, lactate values (24). The results of our study point in the same direction, showing that time-weighted mean lactate, taking every measured lactate value into account, and lactate values at 72 h are better predictors of 90-day survival than the lactate value at admission. We found no independent association between normalization (i.e., time to first normalization of lactate) and 90-day mortality as opposed to several earlier studies. This may in part be explained by different definitions of normalization. Previous studies have defined normalization as percentage of admission lactate, or as timeframes of time to normalization (8 or 24 h frames) used for comparison. Furthermore, the lactate level used to define normalization has also varied, as shown in a recent systematic review (29). The present study showed that a longer duration of hyperlactatemia over 2 mmol/l and a longer time-interval between the first and the last value over 2 mmol/l were associated with higher 90-day mortality. Lactate levels may rise again after temporary normalization or during the course of the illness. Lactate levels are also influenced by liver disease (30) and/or the use of renal replacement therapy (31). Independent association of magnitude, timing, and duration of elevated lactate and 90-day mortality Persistent lactatemia as a continuous and binomial variable, and separately time-weighted mean lactate during ICU stay were both independently associated with 90-day mortality in septic, and separately in septic shock patients. However, time to normalization of hyperlactemia was not, which means that it may not be optimal surrogate outcome measure in future clinical trials. Interaction between lactate levels and use of norepinephrine and epinephrine In this study, all patients had infection and they were hypotensive at the time of study inclusion. However, for some

7 580 SHOCK VOL. 47, No. 5 VARIS ET AL. patients the need for vasopressors was short lasting. Patients needing vasopressors during the first 24 h in the ICU had significantly higher lactate levels than those who did not. Vasopressors may have induced some degree of hypoperfusion and lactatemia by causing peripheral vasoconstriction. Furthermore, epinephrine may cause hyperlactatemia through betaadrenergic stimulation and increased aerobic glycolysis (32). There is also accumulating evidence that the aetiology of hyperlactatemia in sepsis often is multifactorial and beyond tissue hypoxia only (10, 33 35). Nevertheless, even in the absence of tissue hypoxia, hyperlactatemia is indisputably an independent predictor of organ failure (25) and mortality (12, 13) and previous data supporting the clinical utility of lactate as a marker of early sepsis recovery are robust (8, 15, 16). Strengths and limitations Our study has some strengths. First, these data were prospectively collected from consecutive patients admitted to ICUs participating in the FINNAKI study. Second, the 90-day mortality of all septic shock patients in the FINNAKI study was comparable to the mortality of the studied subgroup with multiple lactate measurements. Third, patients were treated by ICU physicians according to the current guidelines without differences in the treatment regimens between patients with different levels of lactatemia. Fourth, we updated this sub-study analysis using the updated definitions for sepsis and septic shock (14). Thus, we consider our study population to be representative and our findings to have external validity. Fifth, this study assessed the association between lactate and a long-term timedefined mortality endpoint with complete outcome data. However, this study also has several limitations. First, due to an observational study design, the lactate measurements were not scheduled, which plausibly may have led to more blood gas samples being drawn from more severely ill patients with higher lactate values. However, the 90-day survivors and non-survivors were sampled comparably. Additionally, we calculated the time-weighted mean lactate, which adjusts for differences in frequency and aggregate time of measurements. Second, a time-dependent variable such as normalization of lactate has the problem of competing risks, such as discharge from the ICU and death. However, imputation of time to death into calculation of normalization of lactate in those who died in the ICU did not change the results. Furthermore, imputation of time to normalization in patients who do not normalize within a certain time also may cause bias. Furthermore, normalization may be followed by a new rise in lactate levels, which may have bearing on outcome. The median ICU LOS was, however, similar among survivors and non-survivors, and, thus, did not cause significant bias to our findings. Fourth, information on source control and appropriateness of antibiotics could not be retrieved from the collected data for assessment of their possible association with evolution of lactate levels. Finally, the selected different sub-populations and use of different regression models may yield different results. However, we could also confirm the robustness of our results in the sensitivity analysis including only patients with lactatemia fulfilling the Sepsis-3 criteria and separately using enter models instead of backward logistic regression models. CONCLUSION In this post hoc analysis of a multicenter observational FINNAKI study, we found that in patients with septic shock and hyperlactatemia (>2 mmol/l), time-weighted mean lactate, lactate values at 72 h, and persisting hyperlactatemia at 72 h were independently associated with 90-day mortality, but time to first normalization of hyperlactatemia and admission lactate were not. These findings may inform the design of future clinical trials using combined surrogate endpoints for mortality, including lactate measurements, in septic shock patients. REFERENCES 1. Fleischmann C, Scherag A, Adhikari NK, Hartog CS, Tsaganos T, Schlattmann P, Angus DC, Reinhart K, and Trialists IFOaC. Assessment of global incidence and mortality of hospital-treated sepsis - current estimates and limitations. Am J Respir Crit Care Med 193: , Adhikari NK, Fowler RA, Bhagwanjee S, Rubenfeld GD: Critical care and the global burden of critical illness in adults. Lancet 376: , Annane D, Aegerter P, Jars-Guincestre MC, Guidet B, and Network CU-R. Current epidemiology of septic shock: the CUB-Rea Network. Am J Respir Crit Care Med 168: , Vincent JL, Sakr Y, Sprung CL, Ranieri VM, Reinhart K, Gerlach H, Moreno R, Carlet J, Le Gall JR, Payen D, and Sepsis Occurrence in Acutely Ill Patients I. Sepsis in European intensive care units: results of the SOAP study. Crit Care Med 34: , Kaukonen KM, Bailey M, Suzuki S, Pilcher D, Bellomo R: Mortality related to severe sepsis and septic shock among critically ill patients in Australia and New Zealand, JAMA 311: , Lagu T, Rothberg MB, Shieh MS, Pekow PS, Steingrub JS, Lindenauer PK: Hospitalizations, costs, and outcomes of severe sepsis in the United States 2003 to Crit Care Med 40: , Rivers E, Nguyen B, Havstad S, Ressler J, Muzzin A, Knoblich B, Peterson E, Tomlanovich M, and Early Goal-Directed Therapy Collaborative G. Early goaldirected therapy in the treatment of severe sepsis and septic shock. N Engl J Med 345: , Arnold RC, Shapiro NI, Jones AE, Schorr C, Pope J, Casner E, Parrillo JE, Dellinger RP, Trzeciak S, and Emergency Medicine Shock Research Network I. Multicenter study of early lactate clearance as a determinant of survival in patients with presumed sepsis. Shock 32:35 39, Sterling SA, Puskarich MA, Shapiro NI, Trzeciak S, Kline JA, Summers RL, Jones AE, and Emergency Medicine Shock Research N. Characteristics and outcomes of patients with vasoplegic versus tissue dysoxic septic shock. Shock 40:11 14, Garcia-Alvarez M, Marik P, Bellomo R: Sepsis-associated hyperlactatemia. Crit Care 18:503, Nichol AD, Egi M, Pettila V, Bellomo R, French C, Hart G, Davies A, Stachowski E, Reade MC, Bailey M, et al.: Relative hyperlactatemia and hospital mortality in critically ill patients: a retrospective multi-centre study. Crit Care 14:R25, Thomas-Rueddel DO, Poidinger B, Weiss M, Bach F, Dey K, Haberle H, Kaisers U, Ruddel H, Schadler D, Scheer C, et al.: Medical Education for Sepsis Source C and Antibiotics Study G. Hyperlactatemia is an independent predictor of mortality and denotes distinct subtypes of severe sepsis and septic shock. J Crit Care 30:439, Casserly B, Phillips GS, Schorr C, Dellinger RP, Townsend SR, Osborn TM, Reinhart K, Selvakumar N, Levy MM: Lactate measurements in sepsis-induced tissue hypoperfusion: results from the Surviving Sepsis Campaign database. Crit Care Med 43: , Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, Bellomo R, Bernard GR, Chiche JD, Coopersmith CM, et al.: The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA 315: , Gu WJ, Zhang Z, Bakker J: Early lactate clearance-guided therapy in patients with sepsis: a meta-analysis with trial sequential analysis of randomized controlled trials. Intensive Care Med 41: , Jansen TC, Van Bommel J, Schoonderbeek FJ, Sleeswijk Visser SJ, Van Der Klooster JM, Lima AP, Willemsen SP, Bakker J, Group LS: Early lactate-guided therapy in intensive care unit patients: a multicenter, open-label, randomized controlled trial. Am J Respir Crit Care Med 182: , 2010.

8 SHOCK MAY 2017 LACTATE AND MORTALITY IN SEPSIS AND SEPTIC SHOCK Nisula S, Kaukonen KM, Vaara ST, Korhonen AM, Poukkanen M, Karlsson S, Haapio M, Inkinen O, Parviainen I, Suojaranta-Ylinen R, et al.:, and Group FS. Incidence, risk factors and 90-day mortality of patients with acute kidney injury in Finnish intensive care units: the FINNAKI study. Intensive Care Med 39: , Bone RC, Balk RA, Cerra FB, Dellinger RP, Fein AM, Knaus WA, Schein RM, Sibbald WJ: Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine. Chest 101: , Bone RC, Sibbald WJ, Sprung CL: The ACCP-SCCM consensus conference on sepsis and organ failure. Chest 101: , Puskarich MA, Trzeciak S, Shapiro NI, Albers AB, Heffner AC, Kline JA, Jones AE: Whole blood lactate kinetics in patients undergoing quantitative resuscitation for severe sepsis and septic shock. Chest 143: , Shankar-Hari M, Phillips GS, Levy ML, Seymour CW, Liu VX, Deutschman CS, Angus DC, Rubenfeld GD, Singer M, Sepsis Definitions Task F. Developing a new definition and assessing new clinical criteria for septic shock: for the third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 315: , Ray P, Le Manach Y, Riou B, Houle TT: Statistical evaluation of a biomarker. Anesthesiology 112: , Wacharasint P, Nakada TA, Boyd JH, Russell JA, Walley KR: Normal-range blood lactate concentration in septic shock is prognostic and predictive. Shock 38:4 10, Nichol A, Bailey M, Egi M, Pettila V, French C, Stachowski E, Reade MC, Cooper DJ, Bellomo R: Dynamic lactate indices as predictors of outcome in critically ill patients. Crit Care 15:R242, Bakker J, Gris P, Coffernils M, Kahn RJ, Vincent JL: Serial blood lactate levels can predict the development of multiple organ failure following septic shock. Am J Surg 171: , Mcnelis J, Marini CP, Jurkiewicz A, Szomstein S, Simms HH, Ritter G, Nathan IM: Prolonged lactate clearance is associated with increased mortality in the surgical intensive care unit. Am J Surg 182: , Nguyen HB, Rivers EP, Knoblich BP, Jacobsen G, Muzzin A, Ressler JA, Tomlanovich MC: Early lactate clearance is associated with improved outcome in severe sepsis and septic shock. Crit Care Med 32: , Song YH, Shin TG, Kang MJ, Sim MS, Jo IJ, Song KJ, Jeong YK: Predicting factors associated with clinical deterioration of sepsis patients with intermediate levels of serum lactate. Shock 38: , Vincent JL, Quintairos ESA, Couto L Jr, Taccone FS: The value of blood lactate kinetics in critically ill patients: a systematic review. Crit Care 20:257, Jeppesen JB, Mortensen C, Bendtsen F, Moller S: Lactate metabolism in chronic liver disease. Scand J Clin Lab Invest 73: , De Corte W, Vuylsteke S, De Waele JJ, Dhondt AW, Decruyenaere J, Vanholder R, Hoste EA: Severe lactic acidosis in critically ill patients with acute kidney injury treated with renal replacement therapy. J Crit Care 29: , Levy B, Desebbe O, Montemont C, Gibot S: Increased aerobic glycolysis through beta2 stimulation is a common mechanism involved in lactate formation during shock states. Shock 30: , Chertoff J, Chisum M, Garcia B, Lascano J: Lactate kinetics in sepsis and septic shock: a review of the literature and rationale for further research. J Intensive Care 3:39, Kraut JA, Madias NE: Lactic acidosis. N Engl J Med 371: , Suistomaa M, Ruokonen E, Kari A, Takala J: Time-pattern of lactate and lactate to pyruvate ratio in the first 24 hours of intensive care emergency admissions. Shock 14:8 12, 2000.