Characteristics and long-term outcomes of patients with lupus-related protein-losing enteropathy: A retrospective study

Patients and data collection

The patients included into this study were admitted to Peking Union Medical College Hospital between January 2000 and June 2016, as described in our previous study,^[16] and they were followed until May 2017. The study was approved by the Medical Ethics Committee of PUMCH. Patients diagnosed with SLE who fulfilled ≥4 of the criteria of the American College of Rheumatology 1997 revised criteria for SLE were enrolled into this study. PLE was diagnosed if a patient had (1) definite PLE, which was based on the clinical symptoms, laboratory test results, and evidence of protein leakage from the gastrointestinal (GI) tract that was confirmed by technetium-99m (^99mTc) human serum albumin (HSA) scintigraphy, or (2) probable PLE, which was defined as the presence of PLE-associated clinical symptoms and laboratory test results that included hypoalbuminemia that could not be completely explained by other pathological mechanisms, including urinary protein loss, severe liver disease, or malnutrition; probable PLE was diagnosed based on the independent judgments of ≥2 experienced rheumatologists. The exclusion criteria included patients who were diagnosed with other autoimmune diseases, including inflammatory bowel disease, which can cause PLE, and the presence of other pathological conditions that could account for a patient's hypoalbuminemia, including lupus nephritis (LN) with massive proteinuria, liver dysfunction, and abnormalities in protein synthesis or protein absorption.

The patients’ baseline data were collected retrospectively from their medical charts including the demographic data, family histories, clinical characteristics, and laboratory test results. The clinical data were collected in accordance with the Systemic Lupus International Collaborating Clinics 2012 criteria.^[18] The laboratory data collected included serum albumin, complement (C) 3, C4, calcium, total cholesterol, and triglyceride levels, and the 24-hour urine protein (24hUPro) levels. Patients’ autoantibody profiles, including the levels of antinuclear antibody (ANA), anti-double-stranded DNA (anti-dsDNA) antibody, and anti-extractable nuclear antigen antibodies, including anti-Sjögren syndrome antigen A (SSA) and anti-Sjögren syndrome antigen B (SSB) antibodies, anti-Smith (Sm) antibody, and anti-ribonucleoprotein (RNP) antibody, as well as the Systemic Lupus Erythematosus Disease Activity Index 2000 (SLEDAI-2K), were obtained from all patients upon their admission to the hospital.

The observation period spanned from the time of initial visit to the time when admitted to our hospital until May 2017. Most patients attended regular follow-up appointments, and their clinical records were reviewed and the follow-up data were collected. A minority of patients attended the follow-up appointments irregularly, and telephone interviews were conducted to gather some of the data. Lost to follow-up was defined as neither attending follow-up appointments at our clinic nor could be reached by telephone.

Definitions of remission and flares

Clinical remission was defined as SLEDAI-2K ≤4 points that was attained with or without the use of glucocorticoid or immunosuppressant therapy. An SLE flare was defined as an increase in the SLEDAI-2K ≥4 points compared with the previous visit, the appearance of a new SLE manifestation, or the worsening of a preexisting clinical or hematological manifestation that led to the resumption of or an increase in the corticosteroid or immunosuppressant therapies after a patient's first remission. An SLE-related PLE flare was defined as a reoccurrence of PLE-associated clinical presentations that resulted in the resumption of or an increase in corticosteroid or immunosuppressant therapy.

Statistical analyses

The data were expressed as the means and the standard deviations (SDs). Graphpad Prism® software, version 6.0 (GraphPad Software Inc., La Jolla, CA, USA) was used to create the Kaplan-Meier survival curves. Demographic, clinical, and laboratory factors were analyzed as potential predictors of flares or death using IBM®SPSS® software, version 21.0 (IBM Corporation, Armonk, NY, USA). The potential risk factors analyzed included involved organs/systems, the SLEDAI, total cholesterol, serum albumin, 24hUPro, anti-dsDNA antibody, anti-Sm antibody, anti-SSA antibody, anti-SSB antibody, anti-RNP antibody, and hypocomplementemia. Kaplan-Meier curves and log-rank tests were used to perform the univariate analyses that determined potential risk factors, and factors with values of P<0.10 in the univariate analyses were included in a stepwise multivariate Cox proportional hazards model. Finally, factors that had values of P≤0.05 were considered statistically significant confounders associated with death or SLE flares. A biomedical statistician undertook a statistical review of the study's data.

Baseline characteristics

Table 1 presents the baseline demographic characteristics of 58 patients who were diagnosed with SLE-related PLE at a median age of 35.3 years (range: 12–71 years). Forty-six patients were female, and the female-to-male ratio was 3.83:1. The mean (SD) time from disease onset to diagnosis was 44.0 (66.1) months. Fifty-three patients presented with symptoms associated with PLE as the initial manifestation of SLE.

Clinical characteristics and laboratory test results

Peripheral edema was present in 52 patients (89.7%), and it was the most common symptom. Almost 2/3 of the patients had GI symptoms, including distension (n = 40, 68.9%), diarrhea (n = 27, 46.6%), vomiting (n = 11, 19.0%), and abdominal pain (n = 10, 17.2%). Fifty-one patients (88.0%) had ascites, 44 (75.8%) had pleural effusion, and 32 (55.2%) had pericardial effusion.

Only 16 patients (27.6%) presented with GI symptoms didn’t have other organ/system involvement, and 42 patients had other organ/system involvement, including serositis (n = 45, 77.6%), LN (n = 34, 57.6%), neuropsychiatric SLE (n = 6, 10.3%), and hematologic disorders (n = 13, 22.4%). Of the 34 patients with LN, 13 (22.4%) had baseline 24hUPro secretions >0.5 g, including 2 with nephrotic syndrome. The mean (SD) SLEDAI-2K score was 7.9 (4.9) points (range: 2–27 points). Regarding the SLEDAI scores, 15 patients (25.86%) had inactive or mildly active (0–4 points) disease, 26 patients (44.8%) had mild (5–9 points) disease, 12 patients (20.7%) had moderate (10–14 points) disease, and 5 patients (8.6%) had severe (≥ 15 points) disease.

The laboratory test results showed that 98.3% of the patients were ANA-positive, 87.9% had hypocomplementemia, 51.7% were anti-SSA antibody-positive, and 62.1% had high total cholesterol levels. These were the most common abnormalities in the laboratory test results. The mean (SD) serum albumin level was 16.61 (4.90) g/L.

Diagnosis and sites of protein leakage

Twenty-six patients (44.8%) underwent ^99mTc HSA scintigraphy, and all had PLE. Most of these patients had proteinuria and intractable hypoalbuminemia, and ^99mTc HSA scintigraphy helped to differentially diagnose the causes of hypoalbuminemia in these patients. The most common site of protein leakage was small intestine in 21 of 26 patients (80.8%), followed by large intestine in 15.4% and the stomach in 3.8% of the patients.

Mortality and survival analysis

Eleven patients were lost to follow-up and excluded; so, 47 patients were finally enrolled in the survival analysis. All 47 patients were followed for >12 months, with an average follow-up duration of 58.6 months.

Five (10.6%) deaths occurred during follow-up. The causes of death were septic shock (n = 2), SLE flare (n = 2), and leukemia (n = 1). The 1-, 3-, and 5-year survival rates were 93.6%, 91.3%, and 88.4%, respectively (Figure 1A).

Figure 1

(A) Kaplan-Meier survival curve for 47 patients with systemic lupus erythematosus (SLE)-related protein-losing enteropathy (PLE). The 1-, 3-, and 5-year survival rates were 93.6%, 91.3%, and 88.4%, respectively. (B) Kaplan-Meier survival curve for patients with SLE-related PLE stratified according to the presence of high levels of urine protein, defined as a 24-hour urine protein (24hUPro) levels ≥0.5 g. The patients with 24hUPro ≥0.5 g had a higher mortality rate (χ² = 4.518, P = 0.033).

Table 2 compares the characteristics of the patients who died and survived. The groups were similar regarding the average serum albumin and total cholesterol levels. Patients who died had a significantly higher mean (SD) 24hUPro level (2.19 [2.70] g) than the patients who survived (0.51 [0.76] g) (P = 0.002).

The Kaplan-Meier analysis, log-rank test, and univariate Cox regression model identified neuropsychiatric involvement, LN, and the 24hUPro level as possible risk factors associated with mortality. The Cox multivariate regression model determined that the 24hUPro level was the only independent risk factor associated with death (β coefficient = 0.440 [0.174], P = 0.012). Figure 1B presents Kaplan-Meier curves that stratified survival according to the presence of high 24hUPro levels that were defined as >0.5 g/24 hours.

Flares and associated risk factors

Five flares were recorded in 47 patients after the first clinical remissions, and all occurred during the processes of immunosuppressant dosage tapering or treatment withdrawal. The 5-year flare rate was 10.64%. Flares manifested as PLE (n = 3), LN (n = 1), or hematological disorder (n = 1). Applying a limit of 1 SD, we stratified the patients into 3 groups according to their serum albumin levels as follows: mild hypoalbuminemia (Alb ≥22 g/L) (10.3%), moderate hypoalbuminemia (12 g/L≤ Alb <22 g/L) (77.6%), and severe hypoalbuminemia (Alb <12 g/L) (12.1%). The Kaplan-Meier analysis, log-rank test, and a Cox regression model confirmed that severe hypoalbuminemia, i.e., a serum albumin concentration <12 g/L, was a risk factor associated with flares (hazard ratio, 10.345; 95% confidence interval, 1.690–63.321) (Figure 2).

Figure 2

Kaplan-Meier curve showing the flare-free rate for patients with systemic lupus erythematosus-related protein-losing enteropathy. Patients with an albumin level ≥0.5 g had higher flare rates (χ² = 9.675, P = 0.002).