Biochemical Markers of Nutrition Support in Critically Ill Trauma Victims

Abstract

Background:

In critically ill patients, plasma serum albumin and transthyretin concentrations are thought to reflect the effects of acute illness, including resuscitation and inflammation. Their use as markers for preexisting nutrition status is, therefore, not recommended. Whether they can be used to assess subsequent effectiveness of artificial nutrition support is unclear. We sought to determine if these biomarkers are associated with enteral caloric intake in critically ill trauma patients.

Materials and Methods:

We analyzed data from adult trauma victims who required ≥2 days of mechanical ventilation and ≥7 days of intensive care. We categorized patients into low, middle, or high enteral calorie delivery groups (2, 9, or 17 kcal/kg/d during the first week). We compared serial concentrations of serum albumin, transthyretin, and C-reactive protein. Multiple linear and Poisson regression were used to determine relationships between calorie intake and nutrition biomarkers.

Results:

In total, 1056 patients were analyzed. Their median age was 44 (interquartile range [IQR], 28–57) years, and median injury severity score was 34 (IQR, 26–41). Calorie intake during the first week was not related to biomarkers during the first or second week. However, by the beginning of the third week, the highest calorie group showed greater changes in concentrations of transthyretin (+3.0 mg/dL relative to initial concentration, P = .01) and serum albumin (+0.17 g/dL, P = .05) compared with the lowest calorie group.

Conclusions:

In trauma patients requiring 1 or more weeks of intensive care, changes in transthyretin were associated with enteral caloric intake. Our data suggest that transthyretin could be used to monitor nutrition support after 2 weeks in intensive care.

Introduction

Inflammation in trauma and critical illness causes changes in vascular permeability and hepatic synthesis, which alter plasma levels of visceral proteins such as serum albumin and transthyretin.¹ In the context of critical illness, changes in these biomarkers over time are thought to reflect gradual resolution of inflammation, rather than reflecting changes in nutrient provision.^2,3 Therefore, traditional nutrition biomarkers have limited interpretation and clinical utility in critically ill populations.^4,5

However, over the first week following injury, the physiology of inflammation gradually diminishes in most patients,¹ and inflammation-related effects on serum albumin and transthyretin may become less influential. For patients with continued need for intensive care and artificial nutrition beyond 1 week, nutrition biomarkers may be less influenced by inflammation, despite ongoing “critical illness.” In the weeks following injury, serum albumin and transthyretin may actually provide valuable information on the effectiveness of calorie and protein delivery.

In this study, we sought to determine the relationship between enteral caloric intake and circulating serum albumin and transthyretin concentrations in critically ill trauma patients requiring more than 1 week of intensive care. We hypothesized that differences in nutrition support during the first week of intensive care unit (ICU) stay would be reflected by differences in nutrition biomarker concentrations in the second and third weeks.

Materials and Methods

Study Design, Patients, and Setting

The data for this study were collected from severely injured trauma patients from 2003–2011 under a prospective research program at our regional level 1 trauma center, as previously described.⁶ Briefly, patients were enrolled in this prospective cohort study if they were adult trauma patients with signs of systemic shock upon admission, defined by (1) a base deficit ≥6 or systolic blood pressure <90 mm Hg and (2) a requirement for any packed red blood cell transfusion. Patients were excluded if they had isolated traumatic brain injury or if they were expected to die within 48 hours of admission. Clinical data were obtained using the electronic medical record. The study was reviewed and approved by the University of Washington Internal Review Board.

Patients were excluded if they required <2 days of mechanical ventilation or stayed in the ICU <7 days. These exclusion criteria were designed to target enrollment to patients who were likely to require nutrition therapy for more than a few days. Per our previously described hospital protocol,⁷ patients were started on parenteral nutrition (PN) only if they did not achieve at least trophic feeding by hospital day 7.

Definition of Calorie Intake Groups, Measurement of Plasma Proteins, and Description of Clinical Outcomes

We wanted to determine whether caloric enteral intake during the first postinjury week was related to transthyretin and serum albumin, measured later during hospitalization. All patients had daily enteral calorie delivery recorded while in the ICU. To test the relationship between caloric intake and nutrition biomarkers, we grouped patients according to their enteral calorie intake during the first week and then compared the various outcomes that are described below.

Main outcome measures were changes in mean plasma concentrations for serum albumin (normal level >3.5 g/dL), transthyretin (normal level >16 mg/dL), and C-reactive protein (CRP, normal level <3 mg/dL) over 49 days. These measurements were obtained as part of routine clinical care. Venous blood was collected in 5-mL K2 EDTA Vacutainer tubes (BD Vacutainer, Franklin Lakes, NJ) and sent to the main hospital laboratory for analysis with automated nephelometric immunoassays (Beckman-Coulter, Brea, CA). Laboratory values for each individual were grouped into time periods: 1–6 days (week 1), 7–14 days (week 2), and 15–21 days (week 3). If patients had multiple values for the same laboratory in 1 time period, the mean of these values was used.

Secondary outcomes included ventilator-free days (28 minus duration of mechanical ventilation; “0” if patient died while receiving mechanical ventilation), ICU length of stay (LOS), hospital LOS, and hospital mortality.

Statistical Analysis and Data Presentation

Categorical data are shown as counts with percentages, and continuous data are shown as medians with interquartile range (IQR). Descriptive analyses were performed for demographic characteristics and clinical outcomes for the entire cohort. The relationships between calorie groups (defined above) and nutrition biomarkers were first summarized using descriptive analyses. Multiple linear regression was used to further examine the association between enteral calorie delivery and nutrition biomarker values. In these regression models, the exposure variable was categories of calorie delivery (previously defined), and the outcome variable was either (a) mean change in laboratory value from week 1 to week 2 or (b) mean change in laboratory value from week 1 to week 3. In addition, a binary outcome variable was created to indicate whether normal range (>16 mg/dL) of transthyretin concentration occurred by day 21. Potential confounding factors in the model included variables for age, injury severity score (ISS), body mass index (BMI), and exploratory laparotomy. We also included whether CRP was ≥100 mg/dL at the time of transthyretin and serum albumin measurements to characterize the acute phase response. Unless indicated otherwise, all presented results are adjusted for these factors.

The relationship between calorie delivery and clinical outcomes (LOS, ventilator-free days, and mortality) was assessed using Poisson regression and linear regression, with the same model structure as described above. Missing data are reported at each time point. To assess the effect of missing data, a supplemental analysis was performed using multiple imputation with chained equations (MICE). All analyses were performed using Stata 12.1 (StataCorp, College Station, TX).

Results

Of 2953 potential patients enrolled, 354 were excluded because they were <18 years old, and 1543 were excluded because they were in the ICU <7 days or required mechanical ventilation for <2 days. This left 1056 patients for analysis (Figure 1). Demographic and clinical information is shown in Table 1 and summarized here. Patients had a median age of 44 (IQR, 28–57), and 281 (27%) were female. Most (924; 88%) sustained a blunt injury, and the median injury severity score (ISS) was 34 (IQR, 26–41). Of note, the 2-kcal group had higher abdominal Abbreviated Injury Score (median score, 2; IQR, 0–4) compared with the 9-kcal and 17-kcal groups (both with median score, 0; IQR, 0–2). The 2-kcal group had 95 patients with laparotomies (36%), compared with 68 patients (19%) in the 9-kcal group and 41 patients (9%) in the 17-kcal group. One hundred patients (10%) started PN within the first week of hospital stay.

Figure 1.

Flow diagram for a cohort study of nutrition therapy and biomarkers of nutrition in trauma patients. ICU, intensive care unit.

Table 1.

Demographic and Clinical Characteristics of the Study Cohort^a.

Characteristic	Total (N = 1056)
Age, y	44 (28–57)
Female sex	281 (27)
BMI, kg/m2	26.7 (23.5–30.0)
Race/ethnicity
Black	50 (5)
Asian	54 (5)
American Indian	26 (3)
Hispanic	77 (8)
White	791 (75)
Mechanism of injury
Blunt	924 (88)
Penetrating	65 (6)
Other	63 (6)
ISS	34 (26–41)
Severe injury (AIS ≥3)
Head	575 (55)
Thorax	632 (60)
Abdomen	308 (29)
Extremity	579 (55)
Parenteral nutrition
Started therapy	124 (12)
Started before hospital day 14	100 (10)
Duration of therapy, d	8 (5–14)
Laparotomy	204 (19)
PRBCs transfusedb	5 (2–10)
Initial base deficit, mEq/L	5.5 (3.0–8.5)
Ventilator-free days	15 (3–20)
ICU LOS	14 (10–23)
Hospital LOS	26 (19–38)
Mortality	127 (12)

AIS, Abbreviated Injury Score; BMI, body mass index; ICU, intensive care unit; ISS, Injury Severity Score; LOS, length of stay; PRBC, packed red blood cell.

^aAdult trauma patients were enrolled in this study if they showed signs of systemic shock on admission, from 2003–2011, at a level 1 trauma center. Patients were excluded if they stayed <7 days in the ICU or required <2 days mechanical ventilation. Categorical data are presented as median (interquartile range), and continuous data are presented as number (percent). Percentages may not add to 100 due to missing values.

^bNumber of units of packed red blood cells.

Summary of the Trends in Enteral Caloric Intake, Serum Albumin, and Transthyretin Over Time

Our data demonstrated that there is marked variation in enteral caloric intake in critically ill patients. Figure 2a illustrates daily enteral caloric intake over the first month of hospitalization. We divided the cohort into 3 groups, based on their average daily caloric intake per kilogram over the first 7 days. The low (mean 2 kcal/kg/d, or 0–44 kcal/kg in the first week), middle (mean 9 kcal/kg/d, or 45–99 kcal/kg in the first week), and high (17 kcal/kg/d, or ≥100 kcal/kg in the first week) groups contained 265, 353, and 438 patients, respectively (Figure 2b).

Figure 2.

(a) Enteral nutrition feeding trends (kcal/kg/d) over 28 days in a cohort of adult trauma patients. (b) Categories of kcal/kg delivery in the first week of hospital admission, with feeding trends shown over 28 days. The dashed lines represent the standard 25-kcal/kg/d recommended intake. Data were obtained from a cohort of adult trauma patients who exhibited signs of systemic shock on admission, stayed ≥7 days in the intensive care unit, and required ≥2 days mechanical ventilation (N = 1056). Patients were enrolled from 2003–2011 at a level 1 trauma center. Box and whisker plots depict median and interquartile range. At the bottom of the graph, total numbers of patients with available kilocalorie data over time are shown.

Plasma serum albumin declined in the days following admission and remained below admission levels at all time points. Serum albumin concentrations remained below normal range (3.5–5.5 g/dL) in 1048 patients (99%) for the entire follow-up period from days 7–49. After trauma, transthyretin levels dropped initially and then gradually increased starting during the second week of admission. Transthyretin measurements remained below normal range (16–40 mg/dL) in 771 patients (73%) for the follow-up period from days 7–49. Median CRP rose rapidly, remained elevated during the first week, and gradually decreased thereafter. CRP measurements remained above normal range (1–3 mg/dL) in 1050 patients (99%) for the follow-up period from days 7–49 (Figure 3).

Figure 3.

Visceral protein trends over 49 days in a cohort of adult trauma patients (N = 1056). Patients were enrolled if they showed signs of systemic shock on admission, from 2003–2011, at a level 1 trauma center. Patients were excluded if they required <7 days intensive care or <2 days mechanical ventilation. Box and whisker plots depict median and interquartile range. At the bottom of the graph, total numbers of patients with available laboratory data over time are shown. CRP, C-reactive protein.

Higher Enteral Caloric Delivery Was Reflected by Higher Transthyretin and Albumin

Prior to day 14, transthyretin and serum albumin were similar among the 3 groups (all P values >.05, Table 2, Supplemental Table S1). By the 21st postinjury day, the highest calorie group had a median transthyretin of 14 mg/dL (IQR, 9.1–20.1) compared with 9.9 mg/dL (IQR, 6.9–15.1) in the lowest calorie group (P = .01). The median serum albumin concentration at day 21 in the highest calorie group was 1.8 g/dL (IQR, 1.4–2.3) compared with 1.5 mg/dL (IQR, 1.1–1.9) in the lowest calorie group (P = .05) (Table 2, Figure 4). Moreover, by day 21, only 34 patients (13%) in the low-calorie group had a normal transthyretin, compared with 85 patients (24%) and 130 patients (30%) in the medium- and high-calorie groups (P = .03 and P < .01, respectively). That is, higher calorie delivery in the first week of admission was associated with higher levels of transthyretin and with more normal-range transthyretin levels by day 21. Serum albumin appeared to be less responsive to calorie intake and remained below admission levels in most patients, regardless of calorie intake.

Table 2.

Relationship Between Calorie Intake, Changes in Transthyretin and Serum Albumin, and Clinical Outcomes in a Cohort of Severely Injured Trauma Patients (N = 1056)^a.

Outcomes	2 kcal/kg/d (n = 265)	9 kcal/kg/d (n = 363)	17 kcal/kg/d (n = 438)	Average Differenceb	P Value	Adjusted P Valuec
Transthyretin, mg/dL
Week 2	7.1 (5.1–10.9)	9.5 (6.6–13.4)	10.4 (6.8–14.9)	+1.2	.13	.17
Week 3	9.9 (6.9–15.1)	12.6 (8.7–19.3)	14.0 (9.1–20.1)	+3.6	<.01	.01
Serum albumin, g/dL
Week 2	1.5 (1.3–1.9)	1.7 (1.4–2.0)	1.7 (1.4–2.1)	−0.1	.70	.88
Week 3	1.5 (1.1–1.9)	1.8 (1.4–2.1)	1.8 (1.4–2.3)	+0.2	.02	.05
CRP, mg/dL
Week 2	162 (97–237)	137 (77–210)	150 (86–235)	+24.8	.09	.29
Week 3	135 (60–214)	79 (32–137)	73 (33–139)	−0.6	.97	.71
Ventilator-free days	12 (0–20)	16 (6–21)	15 (5–20)	+1.4	.05	.21
ICU LOS, d	15 (10–24)	14 (10–21)	15 (11–24)	+0.4	.69	.20
Hospital LOS, d	24 (17–37)	25 (19–35)	27 (19–41)	+1.5	.47	.09
Mortalityd	45 (17)	35 (10)	47 (11)	−37	.02	<.01

CRP, C-reactive protein; ICU, intensive care unit; LOS, length of stay.

^aTrauma patients were enrolled from 2003–2011 at a level 1 trauma center if they required ≥2 days mechanical ventilation and ≥7 days of intensive care. Calorie groups are defined by average daily enteral kcal/kg intake during the first week of admission. Categorical data are presented as median (interquartile range), and continuous data are presented as number (percent).

^bEstimated mean difference comparing the 17-kcal/kg/d group to the 2-kcal/kg/d group.

^cMultiple linear or Poisson regression with robust standard errors was used to assess statistical significance, with model structure as follows: (1) exposure variables = categories of calorie delivery; (2) outcomes = change in laboratory values, ventilator-free days, length of stay, or mortality; and (3) potential confounders (included in adjusted models only) = categorical variables for admission laboratory value, age, injury severity score, body mass index, and exploratory laparotomy. An additional categorical variable for admission week CRP was included as a potential confounder in the models for serum albumin and transthyretin.

^dReported estimate is a relative risk reduction.

Figure 4.

Visceral protein trends over 3 weeks, grouped by categories of kilocalorie delivery in the first week of admission. Data obtained from a cohort of adult trauma patients who exhibited signs of systemic shock on admission, stayed ≥7 days in the intensive care unit, and required ≥2 days mechanical ventilation (N = 1056). Patients were enrolled from 2003–2011 at a level 1 trauma center. Patients were grouped by median enteral kcal/kg/d intake over the first week of hospital stay. Graphed data points depict median laboratory values. Total numbers of patients with available laboratory data are shown below each week. Multiple linear regression with robust standard errors was used to assess statistical significance, with exposure variable: categories of calorie delivery; outcomes: change in laboratory values; and confounders: initial laboratory value, age, injury severity score, body mass index, and exploratory laparotomy. An additional categorical variable for admission week C-reactive protein was included as a potential confounder in the models for serum albumin and transthyretin. *P ≤ .05. **P ≤ .01.

Enteral Caloric Intake and Transthyretin Trends Were Associated With Clinical Outcomes

Clinical outcomes were associated with both enteral caloric intake and with trends in serum albumin and transthyretin. Compared with patients with low calorie intake (2 kcal/kg/d), mid-range calorie intake (9 kcal/kg/d in first week) was associated with 2.6 more ventilator-free days (95% confidence interval [CI], 1.2 to 4.1), 2.1 days shorter ICU LOS (95% CI, −4.2 to −0.1), and a 54% lower mortality (relative risk [RR], 0.46; 95% CI, 0.3 to 0.7). Compared with patients with decreasing serum albumin over 3 weeks, patients with increasing serum albumin had 4.4 more ventilator-free days (95% CI, 2.9 to 5.9), 7.1 days shorter ICU LOS (95% CI, −9.9 to −4.3), and an 81% lower mortality (RR, 0.19; 95% CI, 0.09 to 0.47). Compared with patients with decreasing transthyretin over 3 weeks, patients with increasing transthyretin had 3.6 more ventilator-free days (95% CI, 1.9–5.4) and 50% lower mortality (RR, 0.48; 95% CI, 0.31–0.99) (Supplemental Table S2).

Plasma CRP modified the effect of enteral caloric intake on transthyretin levels. Among patients with CRP <100 mg/dL, transthyretin tended to reflect enteral caloric delivery, but this relationship was not as clear among patients with CRP ≥100 (P < .01) (Figure 5a and 5b).

Figure 5.

The influence of C-reactive protein (CRP) on the relationship between transthyretin and calorie delivery. CRP levels were assayed during week 2 of hospital stay. Data obtained from a cohort of adult trauma patients who exhibited signs of systemic shock on admission stayed ≥7 days in the intensive care unit and required ≥2 days mechanical ventilation (N = 1056). Patients were enrolled from 2003–2011 at a level 1 trauma center. Patients were grouped by median enteral kcal/kg/d intake over the first week of hospital stay Graphs depict median, interquartile range, and maximum/minimum laboratory values.

Supplementary statistical analyses with imputation showed results that were not meaningfully different from those reported above (Supplemental Tables S3 and S4).

Discussion

Systemic inflammation in the early phases of critical injury results in reprioritization of hepatic protein synthesis, with increased production of acute phase proteins, including CRP and haptoglobin, and decreased production of visceral proteins such as serum albumin and transthyretin. Trauma victims often have increased vascular permeability, leading to plasma protein extravasation, which, together with large-volume fluid resuscitations, makes the use of plasma proteins as markers of nutrition status inaccurate. This, at least in part, is the rationale for recommending against the use of serum albumin or transthyretin as markers of nutrition status in critically ill patients.^1–5 Resolution of these inflammation-related processes, rather than caloric intake, is thought to account for most of the change in visceral protein concentrations over time in critically ill patients.³ However, in our study of trauma patients who required >7 days of intensive care with an ongoing acute phase response, the data suggest that changes in these laboratory values actually are associated with preceding enteral caloric intake. Greater enteral caloric delivery in the first week of ICU admission was reflected by greater increases in transthyretin levels by hospital day 21. Moreover, greater enteral caloric delivery was associated with quicker recovery of transthyretin concentrations to normal levels.

Transthyretin (also known as prealbumin) has a half-life of <7 days⁸ and may provide valuable information regarding the effectiveness of enteral intake in critically ill trauma patients in the weeks following injury. However, our data indicate that a laboratory response is not evident earlier than the third week after injury. For this reason, we recommend use of transthyretin preferentially over serum albumin, given its relatively shorter half-life.¹ In fact, serum albumin appears to have a delayed response to nutrition therapy and does not seem to contribute additional information.

Contemporary studies^2,5,8–12 and expert panels⁴ indicate that transthyretin is not a valid biomarker for preexisting nutrition status assessment in the ICU, for all the reasons previously described. Based on this literature, transthyretin is also not considered useful in monitoring the effectiveness of nutrition support. However, there is currently no adequate evidence regarding serial, longer-term assessments with transthyretin in the ICU. Our study analyzed visceral protein concentrations over a relatively long follow-up period and showed that meaningful trends become apparent at 2–3 weeks after admission. Prior studies of critically ill populations likely failed to show responses in transthyretin concentrations due to shorter follow-up periods.^12,13 Our data indicate that, after 1 week in the ICU, trends in transthyretin may characterize subsequent nutrition status and the response to artificial nutrition therapy.

In this study, we showed that both enteral caloric intake and transthyretin levels are linked to ultimate clinical end points. Mid-range calorie intake (mean, 9 kcal/kg/d) in the first week of admission was associated with decreased ICU LOS, more ventilator-free days, and decreased mortality compared with the group with higher calorie intake (mean, 17 kcal/kg/d), which is consistent with prior studies demonstrating either equivalence or superiority of hypocaloric feeding compared with full-goal calorie feeding.^14–17 We also showed that patients with increasing transthyretin experienced more ventilator-free days and shorter ICU LOS, indicating that this marker could also have some prognostic value in the critically ill. While we know that enteral caloric intake and transthyretin levels are associated with outcomes, and we hypothesize that enteral intake affects transthyretin levels, it is not yet known if changing nutrition therapy in response to plasma biomarker levels will affect clinical outcomes. Of note, this is a hypothesis-generating study, and the reported associations between enteral intake and nutrition biomarkers may not be causally linked. Future studies of nutrition interventions in the critically ill should incorporate serial transthyretin assays in their study protocols, which could further elucidate the clinical utility of this biomarker. If our data are reproducible, clinicians may be able to use transthyretin as a marker to titrate enteral and protein support in critically ill patients with a length of stay >2 weeks and specifically target additional supplementation to those patients at higher risk of malnutrition.

Consistent with prior literature,² our data showed that it is reasonable to forego measurement of transthyretin until CRP reaches a level of <100 mg/dL. In our patients who had a CRP persistently above 100 mg/dL, the relationship between enteral caloric intake and transthyretin concentrations was less clear. This finding may indicate that transthyretin has low clinical utility in patients with an ongoing acute phase response but also supports the use of transthyretin in patients after 2 weeks of inpatient stay, when most experience a return of CRP toward normal levels.

There are important limitations to this study. First, it is likely that patients who are less ill tend to be fed sooner and receive more calories in the first week. They likely also have higher levels of visceral proteins because of a less severe inflammatory response. Therefore, our observed associations may be due to severity of injury and illness, rather than calorie intake. However, our adjustments for major indicators of illness, including ISS, surgical status, and CRP level, help address this concern. After these adjustments, associations between calorie intake and transthyretin levels remained. Some residual confounding by “illness severity” may remain, and randomized studies are needed to fully address this. Second, our analyses used multiple methods and tested multiple hypotheses. As a result, a falsely positive association is possible. Third, in this study, calorie intake was used as an overall measure of enteral nutrition (EN) delivery, but it is likely that specific components of EN, like protein and micronutrients, are more impactful for nutrition status than are overall calories.¹⁶ Unfortunately, detailed data on protein and micronutrient delivery were not available in this data set, and future studies are needed to specifically examine these components of EN.

Conclusion

In critically ill trauma patients requiring 1 or more weeks of intensive care, changes in transthyretin at 21 days are associated with preceding enteral caloric intake. These findings suggest that comparing transthyretin concentrations at admission and during the third week can be used to assess nutrition status and possibly guide ongoing therapy in critically ill patients.

Clinical Relevancy Statement.

Serum albumin and transthyretin are poor markers of preexisting nutrition status in critically ill patients. They are also not considered useful in monitoring the effectiveness of nutrition support. However, in our cohort study of critically ill trauma patients requiring more than 1 week of intensive care, transthyretin appeared to be associated with enteral caloric intake. In patients receiving the greatest amount of enteral support during the first week of critical illness, transthyretin increased more rapidly and reached normal levels in many by 3 weeks postinjury. These findings suggest that transthyretin may be useful in monitoring artificial nutrition support.