INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

All organs are able to release lactate in both physiologic and pathophysiologic conditions. The organs producing the largest amounts of lactate are the skin, erythrocytes, skeletal muscles, and leukocytes (20, 16, 12, and 11 mg/kg · min, respectively [1]). The lungs do not release much lactate, so that the arteriovenous differences in lactate across the lungs are small and close to zero under physiologic conditions (2). In critically ill patients, Weil and colleagues (5) observed that venous blood samples from a pulmonary artery catheter yielded lactate concentrations equivalent to those in arterial blood. Similar results were obtained by Nimmo and associates (6) in patients with sepsis or acute respiratory distress syndrome (ARDS). However, Brown and coworkers (7) recently reported in 19 patients that lactate production by the lungs could increase in patients with sepsis and ARDS and that the lung lactate production was proportional to the degree of respiratory failure. Similarly, Douzinas and colleagues (8) recently reported lactate production by the lungs in 14 patients with multiple organ failure including ARDS.

One can wonder if lactate production by the lungs is specific for ARDS or if it could also be observed in other forms of acute respiratory failure requiring mechanical ventilation, such as bronchopneumonia, cardiogenic pulmonary edema, lung transplantation, and other types of respiratory failure. We investigated these diseases to better define the determinants of lactate production by the lungs in ARDS, including inflammation, infection, and ischemia, as well as the degree of extension of these processes (diffuse versus focal). Patients with bronchopneumonia are expected to present with a focal inflammatory process; patients with cardiopulmonary edema, and diffuse noninflammatory process; and patients studied shortly after lung transplantation, a predominant ischemia-reperfusion injury.

Also, some methodologic concerns could be raised since the arteriovenous differences in lactate are usually small, they could be largely influenced by even small errors in lactate measurements. It was therefore essential to base measurements on a precise methodology. Hence, we first evaluated the effects of these errors and how they could be reduced by increasing the numbers of measurements. This methodology could be useful also in other studies evaluating differences in lactate across organs.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Accuracy of Arteriovenous Differences in Lactate

We first analyzed 500 samples, including 194 from critically ill patients with various diagnoses and 306 samples from dogs being used in other studies, during endotoxic shock (9, 10). Of these, 300 single arterial samples were used to determine the coefficient of variation of lactate measurement, and 100 pairs of arterial and venous (mixed-venous and hepatic vein) samples were used to determine the accuracy of arteriovenous differences in lactate (AVLAC).

The methodology was identical for each sample: 4 ml of blood was immediately placed in oxalate fluoride tubes, put on ice, and centrifuged for 10 min at 3,000 rpm. The plasma was then stored at 18° C until analysis. The lactate concentration was determined five times on each arterial sample (2300 STAT plus; Yellow Spring Instruments Inc., Yellow Springs, OH). The range of lactate levels studied was 0.21 to 23.5 (median: 2.1) mEq/L in humans and 0.7 to 14.3 (median: 2.4) mEq/L in dogs. The coefficient of variation of the 300 lactate measurements was 0.94 ± 1.54%.

For determination of the errors in measurements of AVLAC, lactate concentrations were determined 10 times on each of the 100 pairs of arterial and venous samples. The corresponding calculated AVLAC ranged from 0.59 to 2.45 mEq/L. The coefficient of variation of AVLAC obtained using only one lactate measurement was 14 ± 83%. We averaged the first five arterial and five venous lactate measurements to obtain what was considered as the true lactate level, from which the AVLAC was calculated (reference AVLAC). We then used the other five samples to average one to five arterial and venous lactate measurements successively (AVLAC 1 to 5), so as to calculate the bias and agreement between AVLAC 1 to 5 and the referenced AVLAC according to the method of Bland and Altman (11). The data are shown in Table 1. The mean of the absolute values of the referenced AVLAC was used to calculate the relative error in AVLAC in this sample. The bias was negligible. The agreement was 0.224 mEq/L for a single measurement but increased to 0.066 when the number of measurements was increased from one to five. On the basis of these observations, we averaged five arterial and five venous lactate measurements to minimize the relative error in measurements.

Arteriovenous Differences in Lactate in Patients with Acute Respiratory Failure

From December 1993 to December 1995, the study included 122 patients (78 male and 44 female; age: 60 ± 15 yr; Apache II score: 19 ± 9) with acute respiratory failure requiring mechanical ventilation and invasive monitoring. Of these patients, 43 had acute lung injury (ALI) as defined by diffuse lung infiltrates and a Pa_O2/FI_O2 ratio < 300 mm Hg in the presence of a pulmonary artery occluded pressure (PAOP) < 18 mm Hg (12); nine had cardiogenic pulmonary edema due to congestive heart failure (CPE) as defined by lung infiltrates in the presence of a PAOP > 18 mm Hg and known cardiac disease; 37 had severe bronchopneumonia (BPN) as defined by the presence of lobar infiltrates associated with signs of sepsis and a suspected causative microorganism; seven were studied early after single lung transplantation (LTX); and 26 had other causes of respiratory failure (OTHER), including acute exacerbation of chronic obstructive pulmonary disease, lung fibrosis, neurologic disease, or drug overdose. Each patient was monitored with a pulmonary artery catheter (Swan-Ganz catheter 7.5F; Baxter Healthcare, Irvine, CA) and an arterial catheter. All patients were studied within the first 24 h of their episode of acute respiratory failure for which a pulmonary artery catheter was inserted; patients after lung transplantation were studied on admission into the intensive care unit.

The following measurements were obtained: core temperature, arterial pressure, pulmonary artery pressure, PAOP, right atrial pressure, cardiac index, arterial and mixed-venous blood gases, and lactate. Intravascular pressures were obtained at end-expiration. Cardiac index was determined by the thermodilution technique (computer COM 1 or 2; Baxter) using 10-ml aliquots of a cold (< 10° C) solution of 5% dextrose in water and a closed system (CO-set; Baxter). The injection of the bolus started at end-inspiration and a total of five measurements within 10% of each other were averaged. Blood gases were determined (analyzer model ABL3; Radiometer, Copenhagen, Denmark) and hemoglobin saturations measured (OSM3 hemoximeter; Radiometer). Oxygen delivery (DO₂), oxygen consumption (VO₂), venous admixture (Qva/Qt), and indexed pulmonary vascular resistances (PVRI) were calculated using standard formulas. In patients with ARDS, the acute lung injury score proposed by Murray and associates (13) was also calculated.

For lactate measurement, 4 ml of arterial and mixed-venous blood were simultaneously obtained. Lactate concentrations were determined five times on each sample and averaged. Lactate production by the lungs in mEq/min · M² was calculated as the product of AVLAC times the cardiac index.

Differences between groups were analyzed by analysis of variance for repeated measurements. When the F value was significant, a post-hoc adjustment of Neuman-Keuls was applied.

Survival rate and catecholamine use were compared using chi-square analysis. In each group, the correlation between lactate production by the lungs and three other variables was assessed by linear regression.

Statistical significance was accepted for a p value < 0.05. All data are presented as mean ± SD in text and tables, unless stated otherwise.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the entire population, arterial lactate levels ranged from 0.5 to 23.2 mEq/L (mean: 2.6 ± 2.9; median: 1.59). There was no significant difference in arterial lactate levels between the various groups. As expected, patients with ALI had a lower Pa_O2/ FI_O2 ratio and a higher venous admixture than patients with other causes of respiratory failure. Cardiac index was lower in patients with CPE but there were no significant differences in DO₂ between the groups (Table 2). The survival rate was not different among groups.

AVLAC was higher in patients with ALI than in other groups, although the differences did not reach statistical significance for patients with CPE and BPN (Table 2). In patients with ALI, the maximal AVLAC value was 0.85 mEq/L and AVLAC increased in proportion to the lung injury score (0.032 + 0.032x; r = 0.46, p < 0.01) (Figure 1).

View larger version (15K): [in this window] [in a new window]   Figure 1.   Relationship between the arteriovenous difference in lactate (upper panel ) or lung lactate production (lower panel ) and the lung injury score in patients with acute lung injury.

Accordingly, lung lactate production was significantly higher in patients with ALI than in the other groups (Figure 2). In patients with ALI, lung lactate production was inversely related to the Pa_O2/FI_O2 ratio (1.42  0.005x; r = 0.35, p < 0.05) (Figure 3) but directly related to venous admixture (0.36 + 0.003x; r = 0.49, p < 0.01) (Figure 4) and lung injury score (0.19 + 0.36x; r = 0.45, p < 0.01) (Figure 1). Lung lactate production was inversely related to mean arterial pressure (2.17  0.02x; r = 0.34, p < 0.05) but was not significantly correlated with cardiac index, DO₂, VO₂, or PVRI (data not shown). In all groups, lung lactate production was not significantly related to arterial lactate levels.

View larger version (17K): [in this window] [in a new window]   Figure 2.   Lung lactate production in the various groups of respiratory failure (see Figure 3 for abbreviations). Data are presented as mean ± SEM.

View larger version (11K): [in this window] [in a new window]   Figure 3.   Relationship between lung lactate production and the PaO2/FIO2 ratio in the various groups of patients (ALI = acute lung injury; CPE = acute pulmonary edema; BPN = bronchopneumonia; LTx = lung transplantation; OTHER = other types of respiratory failure). *p < 0.05 versus ALI; +p < 0.01 versus ALI.

View larger version (16K): [in this window] [in a new window]   Figure 4.   Relationship between lung lactate production and venous admixture in patients with acute lung injury (closed diamonds) or other types of respiratory failure (open diamonds).

In patients with ALI, lung lactate production was similar in the 30 infected and the 13 noninfected patients (0.76 ± 0.91 versus 0.76 ± 0.98 mEq/min · M²).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study demonstrates that lactate production by the lung can be significantly increased in patients with ARDS and that this increase is proportional to the severity of the lung injury. These results confirm and extend those of Brown and coworkers (7), who observed in 19 patients with ARDS that pulmonary lactate release increased in proportion to the severity of lung disease. Also, Douzinas and colleagues (8) reported an increased AVLAC in 14 patients with ARDS and multiple organ failure. On the other hand, Tagan and associates (14) reported significant lung lactate production in 10 critically ill patients with circulatory failure and an arterial lactate level > 4 mEq/L, whereas it was not significantly different from zero in 38 patients without circulatory failure. In this study, there was no significant relation between lung lactate production and severity of lung disease. In mixed groups of critically ill patients, Weil and colleagues (5) and Nimmo and associates (6) reported that blood lactate levels determined on arterial and mixed-venous samples were similar. However, Weil and colleagues (5) reported an AVLAC as large as 0.82 mEq/L and Nimmo and associates (6) reported an AVLAC of 0.47 mEq/L in some patients, but they did not relate these results to the presence of respiratory failure. Our study indicated that a significant AVLAC can be observed in patients with ARDS and only in these patients.

AVLAC across the lungs are usually small, so errors in measurement could largely influence these gradients. Hence, we first evaluated specifically the influence of errors in measurement of AVLAC. The coefficient of variation for individual lactate determinations was negligible (< 1%) but increased to 14% for AVLAC. Furthermore, the error in AVLAC could be as large as 52% when only one arterial and one venous lactate determination were used to calculate this gradient. Hence, we averaged five arterial and five lactate determinations to limit errors, even when the arteriovenous differences were small. On this basis, one should strongly recommend determining serial lactate samples to study lactate production by an organ. We suggest that at least three arterial and three venous samples should be averaged to determine AVLAC across organs.

AVLAC was somewhat higher in patients with CPE, but this related to the lower cardiac index, so that lung lactate production calculated as AVLAC times cardiac index was not significantly higher than in the other groups of patients without ALI.

Interestingly, lactate production by the lungs was not related to arterial lactate levels. Lung lactate production remained within the physiologic range of lactate production by various tissue beds in healthy subjects (0.6 to 0.8 mEq/kg · h [1]) or in critically ill patients without lactic acidosis (0.8 ± 0.7 [15]), so that lactate clearance by the liver, kidneys, and muscle could easily normalize arterial lactate values. This indicates that the lungs are not the primary source of lactate in critically ill patients.

In normal conditions, the lactate production by the lungs is very small and can only be detected in vitro on tissue slices (16). AVLAC across the lungs is usually small in animals (17- 20) as well as in humans with normal (2- 4). However, this gradient can be increased in various lung diseases, such as granulomatous disease (20), lung carcinoma (4), or ARDS (7). In the present study, the lack of lactate production by the lungs in types of respiratory failure other than ARDS helps to point out the various determinants of lung lactate production in ARDS. First, the inflammatory process seems to be a prerequisite since lactate production by the lungs was significant in patients without any inflammatory process (OTHER) even when pulmonary infiltrates were noted (CPE). The presence of infection was not a prerequisite, since there was no difference in lactate production by the lungs in ALI patients with and without infection. This concept is in agreement with experimental and clinical data: Bellomo and associates (21) reported that the lung released lactate after endotoxin administration in dogs but not in control conditions and Douzinas and colleagues (8) observed in patients with ARDS that lungs released not only lactate but also inflammatory cytokines. Second, the inflammatory process has to be severe. Lactate production was proportional to the severity of hypoxemia and venous admixture. In patients with ALI, lung lactate production was also inversely related to the mean arterial pressure, suggesting that the extent of the inflammatory process plays an important role. Since the decreased mean arterial pressure was not correlated with lung lactate release in patients with CPE, one can assume that the association of a larger release of lactate to a lower blood pressure in patients with ALI reflected a more severe inflammatory state rather than a cause-and-effect relationship related to ischemia. Finally, the process has to be diffuse, since lung lactate production was not significant in localized forms of lung inflammation (BPN and LTX). In localized forms of respiratory disease the arteriovenous difference in lactate originating from the diseased parts of the lungs may be greater, but this excess lactate may be diluted by the effluent of lung zones with normal parenchyma. Rochester and coworkers (4) observed in patients with lung carcinoma that AVLAC was increased only in the diseased lobe so that it remained normal across the whole lung. Lactate may be produced by several mechanisms. Increased lactate production by the lungs may reflect tissue hypoxia. In isolated lungs obtained from normal rats and submitted to various hypoxic gas mixtures, Fisher and Dodia (17) demonstrated that AVLAC was undetectable unless alveolar PO₂ was < 7 mm Hg. Below this value, the lactate/pyruvate ratio also increased. Diseased lungs could be more sensitive to hypoxia. When alveolar PO₂ was lowered to 55 to 65 mm Hg in a dog model of proliferative lung disease obtained by the administration of Freund's adjuvant, Strauss and associates (20) observed that AVLAC and the lactate/pyruvate ratio did increase in the diseased but not in the control lungs. In humans also, Rochester and coworkers (4) observed that the lactate/ pyruvate ratio was unchanged across normal lungs and that this ratio increased from 10.2 ± 1.3 in normal parts of the lungs to 12.2 ± 2.7 in diseased parts. In a recent study available only in an abstract form, Routsi and colleagues (22) observed a significant correlation between AVLAC and the arteriovenous difference in the lactate/pyruvate ratio in critically ill patients. In sepsis, there may be a greater production of lactate even in the absence of hypoxia. Endotoxin may promote lung lactate production probably by inhibition of pyruvate dehydrogenase (23) or by an increased glycolysis (24). Hence, lung lactate production may be due to lung hypoxia or the other cellular alterations. Since we did not measure pyruvate levels, we cannot separate hypoxic from nonhypoxic mechanisms.

Our results do not invalidate those of Weil and colleagues (5) and Nimmo and associates (6), indicating that the site of sampling is unimportant, but stress an important exception, i.e., the presence of ARDS. In patients with ARDS, AVLAC as high as 0.85 mEq/L may be observed. In these patients, venous sampling may more reliably represent abnormalities in cellular metabolism in the other organs. In addition, determination of lactate production by the lungs may be useful to assess the extent of ALI. The present study evaluated lung lactate production only within the first 24 h of the syndrome, but the study of lactate production over time may also be interesting (25).

In conclusion, lung lactate production can be increased in patients with respiratory failure due to ARDS, and this production is proportional to the severity of lung disease.