Protein losing enteropathy and hypoalbuminemia after cardiac surgery

Abstract

Rationale

Postoperative hypoalbuminemia is a common, persistent, and ominous abnormality following major surgery and has uncertain etiology. The purpose of this study was to investigate the incidence of protein-losing enteropathy as measured by alpha-1-antitrypsin clearance after major surgery.

Methods

We conducted a prospective cohort study of consecutive patients undergoing cardiac surgery at one hospital (n = 38, mean age = 65 years, male = 79%). Alpha-1-antitrypsin was measured in both blood and stool samples using standard techniques for diagnosing protein-losing enteropathy. Sensitivity, specificity, and predictive indices were calculated taking into account the measured serum albumin concentration.

Findings

Postoperative hypoalbuminemia was a common finding and occurred in 82% of patients (95% confidence interval: 66 to 91). Paired analyses of blood and stool for alpha-1-antitrypsin clearance showed protein-losing enteropathy in 39% of patients (95% confidence interval: 26 to 55). Stool alpha-1-antitrypsin clearance for detecting postoperative hypoalbuminemia had a sensitivity of 48% (95% confidence interval: 30 to 67) and a specificity of 100% (95% confidence interval: 59 to 100). The positive predictive value was about 100% (95% confidence interval: 59 to 100).

Conclusions

We suggest protein-losing enteropathy may partially explain some cases of postoperative hypoalbuminemia after cardiac surgery.

Introduction

Hypoalbuminemia is defined as a reduction in the circulating protein due to increased losses, decreased production, or acute shifts [1]. Postoperative hypoalbuminemia can be frequent, with 35% to 60% of patients developing this abnormality after major surgery [2–5]. When present, postoperative hypoalbuminemia substantially increases the risk of mortality and morbidity including bacterial pneumonia, respiratory failure, delayed wound healing, surgical site infections, and prolonged hospitalization [5–10]. Moreover, hypoalbuminemia treatment is costly when replacement infusions are initiated that account for over $1 billion in costs globally, are not standard practice, and typically followed by only transient effects [11–13]. Despite the importance, the etiology of postoperative hypoalbuminemia is poorly understood.

Research suggests that occult protein-losing enteropathy may be a common yet unrecognized cause of postoperative hypoalbuminemia. The theory is that perioperative hypotension leads to diversion of blood flow away from the splanchnic micro-circulation, cascades of cellular malfunction over the bowel surface, and a subsequent capillary leak of plasma proteins into the intestinal lumen [14]. This pathophysiology explains several features of postoperative hypoalbuminemia including the concurrent loss of other plasma constituents, failures of parenteral infusions to correct deficits, and an absence of rebound hyperalbuminemia in survivors (unlike reabsorption of fluids from pleural, peritoneal, or other “third spaces”) [15–18]. The theory also accounts for research findings including a reduction of albumin in skin lymphatics, the lack of benefit from albumin infusions, and the absence of observed diarrhea [19–22].

The theory of protein-losing enteropathy contributing to postoperative hypoalbuminemia is also controversial [23]. Some countervailing explanations, for example, have emphasized alternative mechanisms including third space sequestration, acute phase inflammatory reactions, peripheral catabolism, or lymphatic leakage from thoracic duct hypertension [24–28]. The scientific literature agrees the causes of postoperative hypoalbuminemia are incompletely understood with more research needed [29]. To the best of our knowledge, no study has directly tested whether protein-losing enteropathy is a contributor to postoperative hypoalbuminemia. The purpose of this study was to conduct a prospective cohort analysis to examine the incidence of postoperative hypoalbuminemia and occult protein-losing enteropathy in patients undergoing major surgery.

Methods

Patients and setting

We included consecutive patients undergoing cardiac surgery, chosen because it is a major operation unrelated to the gastrointestinal tract yet often complicated by postoperative hypoalbuminemia. We recruited patients at one large teaching hospital identified because all investigators were active staff, had clinical contacts, and could be available to facilitate specimen collection. The recruitment interval spanned from March 18, 2010 to April 13, 2010, selected because it was research time for one author (AMP). We excluded patients who declined consent or who were unable to provide a stool sample. We also replicated analyses on random samples from two investigators (DAR, AMP) who served as pilot data for validation in this study and not reported in results.

Clinical characteristics

We recorded each patient’s age, gender, height, weight, past medical history, surgical details, preoperative blood tests, postoperative blood tests, renal function, liver function, medication treatment, and postoperative complications. Serum albumin was measured using a clinical laboratory MODULAR analyzer with an ALB Plus reagent kit by Roche Diagnostics, Montreal, Quebec. The reagent was based on the standard bromocresol green colorimetric assay with a reference range from 35 to 55 g/L; therefore, hypoalbuminemia was defined as a measured serum albumin < 35 g/L. Between-day coefficients of variation of this assay were 2.48% at 26 g/L and 10.53% at 43 g/L, respectively. All data were obtained in a manner blind to subsequent stool studies.

Stool samples

We identified the first postoperative bowel movement on each patient, recorded the total volume, and collected up to a maximum of 100 g in a plastic container. We removed 2-to-3-g aliquot from each sample, added 10 mL of reagent grade water and vortexed the mixture to yield a homogenous stool suspension. The suspension was then frozen and stored at -20 degrees Celsius to allow batch processing on May 13, 2010 (range: 23–56 days delay from original collection). For analysis, each stool sample was lyophilized, re-suspended in ammonium hydroxide buffer and assayed with a Behring Nepholometric II analyzer and N Antiserum to human alpha-1-antitrypsin reagent by Siemens Health Diagnostics Inc, Newark, Delaware.

Serum A1AT concentration

Serum alpha-1-antitrypsin (A1AT) concentration was assayed using a stored blood sample obtained within 3 days following surgery. These paired samples of stool and blood were then sent in batches to a reference laboratory for analysis (Hospital for Sick Children, Toronto, Canada). The paired serum A1AT sample was analyzed with the Integra 400 analyzer and Tina-quant alpha-1-antitrypsin version 2 reagent by Roche Diagnostics, Montreal, Quebec. Serum A1AT concentration assays were conducted blind to the results of stool A1AT assays, stool A1AT assays were conducted blind to the results of serum A1AT concentration assays, and both A1AT assays were conducted blind to the patient’s clinical data.

Protein loss

We calculated A1AT clearance, the accepted measure of protein-losing enteropathy [30]. Similar to measuring creatinine clearance to assess renal function, each patient’s A1AT clearance was estimated from the patient’s A1AT concentration in stool (stool_A1AT), A1AT concentration in blood (blood_A1AT), and total bowel movement size (stool_volume). The A1AT clearance was calculated based on the Filtration Rate Equation (clearance = stool_A1AT x stool_volume / blood_A1AT). Results were expressed as milliliters per day. The reference range at the central laboratory was less than 22 mL per day, where greater values indicate higher amounts of protein-losing enteropathy. We did not conduct injected radio-labeled albumin assays or other different ways to diagnose protein losing enteropathy [31–33].

Statistical analysis

The main comparison evaluated the frequency of an abnormal A1AT clearance and contrasted patients with postoperative hypoalbuminemia to patients without postoperative hypoalbuminemia. The primary analysis was based on Fisher’s exact test. Secondary analyses estimated sensitivity, specificity, positive predictive value, and negative predictive of abnormal A1AT clearance according to standard epidemiology calculations for a diagnostic test and based on exact binomial 95% confidence intervals [34]. Additional tests evaluated the stool A1AT concentration because it is another measure of protein-losing enteropathy that is more convenient in clinical practice. The sample size was designed to provide sufficient power to detect a correlation coefficient of 0.50 or greater.

Results

Descriptive overview

We screened 45 eligible patients, of whom 43 agreed to participate. A total of 5 individuals were excluded due to missed collection (n = 2), sample spillage in transport (n = 2) or no bowel movement while in hospital (n = 1). A total of 7 patients missed specimens of their first postoperative bowel movement but were retained in analyses by providing a subsequent stool specimen. This yielded a total sample size of 38 patients (Table 1). Overall, half the surgeries were conducted on an emergency basis, the mean duration of anesthesia was 5.6 h, and one-third also involved a second procedure beyond coronary artery grafting (most commonly an aortic valve replacement). The median stool sample was obtained on day 4 following surgery and the mean length of stay was 9.6 days in hospital.

Table 1.

Patient characteristics

		Hypoalbuminemia Following Surgery (n = 31)	Normal Albumin Following Surgery (n = 7)
Demographics
Age < 65 years		17 (55)	6 (86)
Age ≥ 65 years		14 (45)	1 (14)
Male gender		23 (74)	7 (100)
Female gender		8 (26)	0 (0)
Height < 1.7 m		13 (42)	0 (0)
Height ≥ 1.7 m		17 (55)	7 (100)
Weight < 70 kg		10 (32)	1 (14)
Weight ≥ 70 kg		20 (65)	6 (86)
BMI* < 25		10 (32)	2 (29)
BMI* ≥ 25		20 (65)	5 (71)
Past history
Hypertension		23 (74)	3 (43)
Diabetes		11 (35)	2 (29)
Smoking ongoing		4 (13)	0 (0)
Cerebrovascular disease		6 (19)	0 (0)
Peripheral vascular disease		3 (10)	1 (14)
Renal disease		9 (29)	0 (0)
Liver disease		1 (3)	0 (0)
GI disease †		8 (26)	1 (14)
Arthritis		3 (10)	1 (14)
Depression		2 (6)	1 (14)

Data are mean (percentage) of corresponding group

Information on height and weight missing for one patient

* Body Mass Index is weight (kg) divided by square of height (m)

† Denotes Gastrointestinal disease

Serum albumin

The mean serum albumin prior to surgery was 42 g/L (range: 31 to 49, median: 42). The mean serum albumin following surgery was 30 g/L (range: 17 to 45, median: 30). No patient had an increase in serum albumin following surgery (Fig. 1). The smallest individual fall in serum albumin was 0 g/L, the greatest individual fall in serum albumin was 20 g/L. In total, 31 patients developed postoperative hypoalbuminemia and 7 patients did not develop postoperative hypoalbuminemia. This was equivalent to an incidence rate of 82% (95% confidence interval: 66 to 91). As expected, both of the two control samples showed no hypoalbuminemia.

Fig. 1.

Broken ladder plot of serum albumin measured on individual patients before and after surgery. X-axis shows time as marked by preoperative sample and postoperative sample. Y-axis shows serum albumin concentration with a lower limit of normal of 35 g per liter. Dashed horizontal line shows sectors of normal and abnormal albumin. Colored diagonal lines indicate values of specific individuals. Overall results show the majority of patients have a normal serum albumin before surgery, a low serum albumin after surgery, and only one had no decrease in serum albumin

Differences in clinical characteristics

Patients who developed postoperative hypoalbuminemia had a greater mean age and burden of comorbidities compared to patients who did not develop postoperative hypoalbuminemia (Table 1). Surgical procedures were also more extensive among those patients who subsequently developed postoperative hypoalbuminemia (Table 2). Baseline and follow-up electrolytes and other common blood test results were similar in the two groups (Table 3). No patients received intravenous albumin perioperatively. No class of medications was disproportionately represented among those patients who subsequently developed postoperative hypoalbuminemia (Table 4).

Table 2.

Surgery details

			Hypoalbuminemia following surgery (n = 31)	Normal albumin following surgery (n = 7)
Emergency procedure			15 (48)	1 (14)
Duration of operation ≥ 6 h			17 (55)	1 (14)
Total pump time ≥ 2 h			16 (52)	2 (29)
Number of bypass grafts ≥ 3			16(52)	1 (14)
Additional procedure *			12 (39)	1 (14)
Any complication †			5 (16)	0 (0)

Data are mean (percentage) of corresponding group

Information on height and weight missing for one patient

* Includes aortic valve replacement, mitral valve replacement, tricuspid valve repair, aorta repair

† Includes paraplegia, major hemorrhage, or tamponade

requiring reoperation (n = 1 for each)

Table 3.

Laboratory test results

			Hypoalbuminemia following surgery (n = 31)	Normal albumin following surgery (n = 7)
Before surgery
Sodium [mmol/L]			139 ± 3	140 ± 3
Potassium [mmol/L]			3.9 ± 0.5	4.3 ± 0.3
Chloride [mmol/L]			104 ± 4	106 ± 4
Bicarbonate [mmol/L]			26 ± 3	26 ± 3
Creatinine [μmol/L]			115 ± 10	80 ± 6
Urea Nitrogen [mmol/L]			8.0 ± 3.8	5.6 ± 0.8
Bilirubin [μmol/L]			10 ± 6	9 ± 3
AST [U/L] *			32 ± 24	26 ± 6
ALT [U/L] †			38 ± 33	34 ± 10
Hemoglobin [g/L]			135 ± 18	146 ± 7
After surgery
Sodium [mmol/L]			140 ± 4	138 ± 2
Potassium [mmol/L]			4.4 ± 0.5	4.3 ± 0.5
Chloride [mmol/L]			109 ± 6	107 ± 2
Bicarbonate [mmol/L]			23 ± 3	24 ± 3
Creatinine [μmol/L]			114 ± 90	85 ± 23
Urea Nitrogen [mmol/L]			10.5 ± 14	6.8 ± 1.7
Bilirubin [μmol/L]			13 ± 10	13 ± 7
AST [U/L] *			70 ± 58	35 ± 7
ALT [U/L] †			34 ± 40	23 ± 2
Hemoglobin [g/L]			94 ± 12	119 ± 23

Data are mean (± standard deviation) of corresponding group

No patient developed liver failure or kidney failure

* Denotes aspartate aminotransferase

† Denotes alanine transaminase

Table 4.

Postoperative course

			Hypoalbuminemia following surgery (n = 31)	Normal albumin following surgery (n = 7)
Medications
Anti-platelet agent			24 (77)	4 (57)
Oral anti-coagulant			6 (19)	0 (0)
Cholesterol reducing agent			20 (65)	4 (57)
ACE or ARB *			19 (61)	3 (43)
Beta blocker			16 (52)	2 (29)
Calcium channel blocker			10 (32)	1 (14)
Digoxin			3 (10)	0 (0)
Diuretic			11 (35)	0 (0)
Proton pump inhibitor			10 (32)	2 (29)
Diabetes medication †			9 (29)	1 (14)
Complications
Anorexia			9 (29)	1 (14)
Nausea			9 (29)	1 (14)
Vomiting			5 (16)	1 (14)
Worsening pedal edema			11 (35)	0 (0)
Prolonged hypotension			4 (13)	0 (0)
Delirium			3 (10)	0 (0)

Data are mean (percentage) of corresponding group

* Denotes Angiotensin Converting Enzyme inhibitor or Angiotension Receptor Blocker

† Denotes injection or oral treatments

Differences in other laboratory tests

Clinical monitoring following surgery provided no readily available explanation for postoperative hypoalbuminemia. Few patients had postoperative vomiting with no significant differences between the two groups. No patient developed liver failure (as measured by a bilirubin twice the normal range) with no significant differences in lesser changes in bilirubin between the two groups. No patients developed renal failure (as measured by doubling of serum creatinine) with no significant differences in lesser changes of creatinine between the two groups. In addition, we found no major evidence of hyponatremia, hypokalemia, significant bleeding, or other explanation to account for the frequency of postoperative hypoalbuminemia.

Clearance of A1AT

The mean stool A1AT concentration following surgery was 1.078 mg / g dry weight (range: 0.064 to 5.161). In total, 13 patients had a stool A1AT concentration above 1.500 mg / g dry weight and 8 patients had a stool A1AT concentration above 2.000 mg / g dry weight. The mean serum A1AT concentration following surgery was 2.58 g / L (range: 1.19 to 3.47). The mean stool volume was 233 mL (range 40 to 600). In total, 15 patients had an A1AT clearance above 22 mL per day, equal to a 39% overall frequency of protein losing enteropathy (95% confidence interval: 26 to 55). A total of 11 patients had a clearance above 30 mL per day, 7 patients had a clearance above 50 mL per day, and 1 patient had a clearance exceeding 100 mL per day.

Sensitivity and specificity

Patients who developed postoperative hypoalbuminemia were significantly more likely to have an abnormal A1AT clearance than patients who did not develop postoperative hypoalbuminemia (47% vs 0%, p = 0.029). The specificity was 100% (95% confidence interval: 59 to 100). The sensitivity was 48% (95% confidence interval: 30 to 67). The positive predictive value was 100% (95% confidence interval: 78 to 100). The negative predictive value was 30% (95% confidence interval: 13 to 53). At the extremes, greater abnormalities in A1AT clearance correlated poorly with severity of postoperative hypoalbuminemia (Fig. 2).

Fig. 2.

Scatterplot of serum albumin related to protein losing enteropathy as measured by alpha-1-antitrypsin clearance in patients after surgery. X-axis shows clearance expressed as milliliters per day (upper limit of normal = 22 ml per day). Y-axis shows postoperative serum albumin concentration (lower limit of normal = 35 g per liter). Dashed vertical and horizontal lines show sectors of normal and abnormal albumin along with normal and abnormal alpha-1-antitrypsin clearance. Solid diagonal line indicates estimate from ordinary least squares regression across the full range of values. Correlation coefficient based on Spearman statistic

Stool A1AT concentration rather than clearance

Analyses based on stool A1AT concentration rather than A1AT clearance were less predictive of postoperative hypoalbuminemia. Based on a threshold of 1.500 mg / g dry weight as an upper-limit of normal, the specificity was 71%, the sensitivity was 36%, the positive predicted value was 85%, and the negative predictive value was 20%. Based on a threshold of 2.000 mg / g dry weight as an upper-limit of normal, the specificity was 100%, the sensitivity was 26%, the positive predicted value was 100%, and the negative predictive value was 23%. At the extremes, stool A1AT concentration did not significantly correlate with a greater severity of postoperative hypoalbuminemia (r = 0.22, p = 0.179).

Discussion

We studied patients undergoing cardiac surgery at one large hospital. We found postoperative hypoalbuminemia occurred in most cases, with an 82% overall incidence (95% confidence interval: 66 to 91). We also found that postoperative protein-losing enteropathy was also common, with a 39% overall observed frequency (95% confidence interval: 26 to 55). An abnormal A1AT clearance rate was a specific but insensitive indicator of postoperative hypoalbuminemia. No patient with a normal postoperative albumin showed an abnormal A1AT clearance yet the degree of abnormality in A1AT clearance correlated weakly with the patient’s severity of hypoalbuminemia. Together, these data suggest that occult protein-losing enteropathy partially explains some cases of postoperative hypoalbuminemia in patients following surgery.

This is the first study to examine protein-losing enteropathy in the immediate postoperative setting. Most past research has used the assay to monitor patients with chronic inflammatory bowel disorders and generally observed clearance rates ranging from 45 to 220 mL per day [35–41]. A case series of patients referred for cryptogenic gastrointestinal disorders with unexplained hypoalbuminemia found that about half had abnormal A1AT clearance rates above 60 mL per day [42]. Other past studies have examined stool A1AT concentrations rather than clearance rates and found abnormalities among patients with HIV associated colitis, graft vs host disease, cow’s milk allergy, colon cancer, severe burns, and following intense exercise [43–50]. Together, these data suggest that protein-losing enteropathy may be more common in complicated patients than is generally recognized.

Our results suggest that protein-losing enteropathy contributes to postoperative hypoalbuminemia but does not predict the severity of hypoalbuminemia. First, many patients who developed hypoalbuminemia showed no abnormal A1AT clearance or abnormal stool A1AT concentration. Second, the highest observed A1AT clearance (113 mL per day) might account for a decrease in serum albumin to 30 g/L after two days; however, hypoalbuminemia was often more profound in our patients after surgery. Third, the patient with the greatest degree of hypoalbuminemia was not the patient with the highest A1AT clearance. More research is needed to integrate all sources of production and loss including losses at the surgical site, perioperative hypotension with liver ischemia, thoracic duct hypertension with intestinal lymphangiectasia, the time profile of protein losing enteropathy, and the potential for radio-labeled albumin studies.

Our study has several limitations that merit emphasis. The data reflect experience at a single center with a small sample size and no repeated measures to test how stool and serum concentrations change over time. Delays also occurred with specimen processing and bowel transit time; hence, the analytic results may underestimate stool A1AT losses due to artifacts from degradation, lagging changes, incomplete sampling, and false negative immunological tests [35, 51, 52]. The stool assay, itself, is an imperfect measure and misses 27% to 45% of intestinal protein loss when compared to radiolabeled albumin methods [38, 53–55]. We also have no data on ESR, CRP, procalcitonin or other markers of inflammation. Together, these limitations suggest that future studies of patients with postoperative hypoalbuminemia are justified since our study may underestimate the contribution of protein-losing enteropathy and our estimates have wide 95% confidence intervals.

Our research also highlighted the advantages and disadvantages of stool studies in the postoperative setting. We found that patients were remarkably cooperative with the study including holding a positive toward the idea of a scientific study of stool. Clinical staff, in contrast, were somewhat averse to specimen collection and handling. In addition, bowel movements are intermittent and infrequent so that practical methods are necessary for collecting fresh samples during evening, nights, and holidays (unlike urine samples that are more frequently available for measuring and monitoring nephrotic syndrome). Furthermore, the common complication of postoperative ileus means that a formal 72-h stool collection may be a more reliable method than a single stool sample for measuring the severity of protein losing enteropathy (albeit with less clinical feasibility) [56, 57].

Hypoalbuminemia is not limited to the postoperative setting and is also a common finding in many other critical illnesses including sepsis, trauma, burns, and cardiac shock [58, 59]. Our study is the first to show a contribution from occult protein-losing enteropathy toward postoperative hypoalbuminemia. The research also describes the feasibility of stool studies in this setting and basic methods for interpreting results. The positive results also help explain why coliforms have ample substrate to proliferate in patients following major surgery. Additionally, the potential translocation of antigens from the alimentary canal into the circulation may also contribute to a systemic inflammatory syndrome [60]. Stool studies for A1AT clearance can be a safe, available, and feasible method for diagnosing some cases hypoalbuminemia following major surgery.

Summary box

What is already known?

• hypoalbuminemia is a frequent abnormality after major surgery.

• hypoalbuminemia is also common in trauma patients or sepsis syndromes.

• hypoalbuminemia is associated with increased mortality and morbidity.

• the pathophysiology of hypoalbuminemia is poorly understood.

• direct treatment of hypoalbuminemia is often unsuccessful.

What are the new findings?

• hypoalbuminemia is frequent in postoperative cardiac surgery patients.

• protein-losing enteropathy is present in one third of such cases.

• stool alpha-1-antitrypsin clearance in this setting can detect protein losing enteropathy with high specificity yet low sensitivity.

What are possible implications?

• protein losing enteropathy may explain some cases of hypoalbuminemia after major surgery and other critical illnesses.

• stool alpha-1-antitrypsin clearance can be considered for diagnosing patients with post-operative hypoalbuminemia.

• an understanding of protein losing enteropathy might explain why hypoalbuminemia is a marker but not the sole cause of mortality and morbidity.

Author contributions

The lead author (DAR) wrote the first draft. All authors (DAR, CMC, AAP) contributed to study design, manuscript preparation, data analysis, results interpretation, critical revisions, and final decision to submit. The lead author (DAR) and final author (AMP) had full access to all the data in the study, take responsibility for the integrity of the data, and are accountable for the accuracy of the analysis. The lead author acted as guarantor (DAR).

Funding

This project was supported by the Canada Research Chair in Medical Decision Sciences, Canadian Institutes of Health Research, the Kimmel Schatzy Research Fund, the Ornge Air Ambluance Organization and Department of Medicine Resident Research Program.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The research methods involving patients were approved by the Research Ethics Board of Sunnybrook Health Sciences Centre. The lead author (DAR) had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the analysis.

Disclosures

The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript. All authors have no financial or personal affiliations that could influence the decisions and work on this manuscript.

Competing interests

The authors declare no competing interests.