Abstract
Introduction:
Gastrointestinal (GI) complications including graft versus host disease (GVHD) are a major cause of morbidity and mortality in allogenic stem transplant recipients. While several studies have previously looked into the acute gastrointestinal complications, fewer smaller studies have reported late complications. In this large study, we focus on the late (100 days post-transplant) gastrointestinal complications in allogenic stem transplant recipients
Methods:
This is a single center, retrospective study, of all of the adult allogenic stem cell transplant recipients who had their transplant at Duke University over a six year period. A total of 479 patients underwent allogenic stem cell transplant, of whom 392 recipients survived for at least 100 days post-transplant. Late gastrointestinal symptoms were noted in 71 patients, prompting endoscopic evaluation. The primary end point of our study was the diagnosis of GI-GVHD based on endoscopic findings, while overall survival and non-relapse mortality were the secondary end points.
Results:
Of the 71 patients who underwent endoscopy, 45 (63%) patients had GI-GVHD. Of these 45 patients, 39 (87%) had late acute GVHD, 1 (2%) had chronic GVHD, and 5 (11%) patients had overlap disease. Of the patients who did not have GVHD, the symptoms were mostly related to infectious and inflammatory causes. Less common causes included drug toxicity, food intolerance, disease relapse, and motility issues. In a multivariate analysis, the factors that were most indicative of GI-GVHD were histological findings of apoptosis on the tissue specimen (OR=2.35, 95% CI=1.18 – 4.70, p=0.015) and clinical findings of diarrhea (OR=5.43, 95% CI =1.25 – 23.54, p=0.024). The median survival time from the first endoscopy was 8.5 months. The incidence of non-relapse mortality at 6 months was 31% in patients with GI-GVHD and 19% in patients without GI-GVHD (p=0.42). All the patients with GI-GVHD were on steroid therapy and 31% of them received total parental nutrition.
Conclusions:
In our population, close to one-fifth of allogenic transplant recipients experienced late GI complications, warranting endoscopic evaluation. Most of these patients were found to have GI-GVHD which had a high incidence of non-relapse mortality at 6 months and close to one-third of these patients needed total parental nutrition.
Introduction
Gastrointestinal (GI) complications are a major cause of morbidity and mortality after allogeneic hematopoietic stem cell transplantation (HSCT).(1, 2) The gut may be affected at any point during HSCT, from pre-transplant conditioning (e.g., chemotherapy and radiation-induced nausea, vomiting, and diarrhea; mucositis) to soon after engraftment (e.g., bacterial infections, acute graft-versus-host disease (GVHD)) to late recovery (e.g., viral infections, late acute or chronic GVHD). There are several well-written reviews and textbook chapters describing these problems.(3, 4) However, most studies focus on early complications; those that discuss late complications are limited in size (5) or focus solely on chronic GVHD.(6)
The goal of this study was to better characterize and understand late GI complications after allogeneic HSCT. Because GVHD is one of the major causes of post-HCT morbidity and mortality, affecting 40–70% of patients (7, 8), we focused on GI-GVHD while also examining other causes of morbidity. We defined “late” as more than 100 days after allogeneic transplant. Although 100 days is an arbitrary landmark, it remains commonly used in clinical practice since most acute toxicities associated with pre-transplant conditioning have resolved by this time and most patients would have been discharged back to their local physicians. The study objectives were to determine the frequency of GVHD versus non-GVHD GI complications, to estimate the overall survival (OS) and non-relapse mortality (NRM) for patients with late GI complications, to assess the relationship between the clinical diagnosis of GVHD with OS and NRM after adjusting for demographic and transplant variables, symptoms at endoscopy, pathologic features of biopsy, and to identify variables predictive of GVHD. To address these objectives, we performed a single institution, retrospective review of all adult patients who underwent endoscopy and biopsy for GI symptoms >100 days post-transplant. To our knowledge, this is the largest series focusing exclusively on late GI complications following allogeneic HSCT.
Methods
Patients
This study included all adult patients (ages 18 years or older) who had received allografts at Duke University Medical Center over a 6 year period and subsequently underwent esophagogastroduodenoscopy and/or flexible sigmoidoscopy or colonoscopy for GI symptoms. As a retrospective study, we reviewed all endoscopies performed >100 days after HCT. The 71 patients who were > 100 days post-transplant at the time of endoscopy are reported here. As a reflection of real-world experience, there was not a strict rule to warrant endoscopy; rather, this was a clinical decision by the treating physician at the time based on the presence of GI symptoms and uncertainty as to the etiology of those symptoms. We did exclude endoscopies done in the absence of GI symptoms (e.g. we excluded screening colonoscopies), and it is not our clinical practice to perform endoscopies for mild GI symptoms. Only cases in which the GI symptoms were considered treatment-related were included. In the event of multiple endoscopic procedures, only data surrounding the first endoscopy was considered so as to count each patient only one time. Endoscopic findings and complications were reported separately (9) and not included here, though histological findings were re-examined as described below and included.
Data
Investigators used the Duke Adult Bone Marrow Transplant Database to identify target patients and extract relevant data. Variables of interest included (1) patient demographics and transplant-related factors (age at transplant, gender, race, primary disease, donor type, cell type, conditioning regimen, baseline platelet count and bilirubin, history of acute GVHD and chronic GVHD); (2) presenting signs and symptoms (anorexia, abdominal pain or cramping, nausea, vomiting, diarrhea); (3) pathologic and histologic features at endoscopy; and (4) clinical diagnosis. The primary objective was to distinguish GI-GVHD from non-GVHD complications; in cases of non-GVHD complications, we determined the etiology of symptoms based on chart review.
Charts were reviewed and the diagnosis of GVHD was confirmed per NIH Consensus Criteria and graded according to IBMTR criteria (10). While the consensus guidelines do not provide definitions of persistence or recurrence of acute gut GVHD in this cohort of 39 individuals with late onset acute gut GVHD, 27 subjects had a denovo onset without prior acute enteric GVHD. The other 12 had prior acute GI GVHD which had completely resolved. This quiescent onset of late gut acute GVHD occurred a median of 96 (range, 25–597) days after resolution of prior acute gut GVHD. Diarrhea, anorexia, nausea, and vomiting were recorded if there was as any mention of these symptoms to the charts. Two expert GI pathologists (DMC and MJS) reviewed the original biopsies blind to the prior pathology diagnosis and clinical, endoscopic, and laboratory findings. Using the 2014 NIH consensus guidelines (11, 12), a consensus diagnosis was made for each individual biopsy site (not GVHD, possible GVHD or likely GVHD). Additionally, various histologic features were quantitatively (e.g., eosinophils per high power field) or qualitatively (e.g., apoptosis as 0=none, 1=minimal, 2=mild, 3=moderate, 4=marked/severe) evaluated. The later data points included intraepithelial apoptosis, gland/crypt destruction, gland/crypt drop out, gland/crypt atrophy, lamina propria (LP) fibrosis/hyalinization, LP inflammation, cryptitis, and ulceration. Biomarkers such as Reg3 and ST2 were not assayed and not included in analysis. The etiology of GI symptoms was determined based on a combination of pathologic review and clinical criteria (e.g. treatment and/or response to glucocorticoids, antibiotics, etc.).
Endpoints
The primary endpoint was the diagnosis of GVHD; secondary endpoints included OS and NRM. OS was defined as the interval from the date of endoscopy until death due to any cause or last contact for surviving patients. OS was estimated using the Kaplan-Meier product limit method. NRM was defined as the interval from the date of endoscopy until death in the absence of disease relapse. The event was summarized by the cumulative incidence estimate with disease-related death as the competing risk; surviving patients were censored at the date of last contact.
Statistical Analysis
To identify factors related to GVHD, we used logistic regression in which GVHD (positive or negative) was the dependent variable. We first built three separate multivariate logistic models with candidate variables from each category: (1) demographic and disease/transplant-related, (2) symptoms and endoscopy-related, and (3) histologic features, respectively. Variables were selected using forward stepping with 0.10 as the cutoff significance level. Odds ratios (OR) and their 95% confidence intervals (CI) were taken from respective models. Time to event variables was estimated according to the product limit method; comparisons between groups of such variables used the logrank test. To assess the relationship between various independent variables and OS, proportional hazards regression was used. Hazard ratios (HR) and 95% CIs were taken from respective models. Comparison of two or more proportions used a chi square test or Fisher’s exact test in the case of small numbers. Comparisons of continuous data across two or more groups used the Kruskal-Wallis test.
Cumulative incidence graphics and tests between groups in the presence of competing risks used R 3.1.1; Kaplan-Meier graphics used S+ 8.1. Logistic and proportional hazards regressions used SAS 9.2.
Results
During the six-year study period, 1134 patients underwent HSCT; 479 received allogenic transplants, of whom 392 survived through at least day 100. Of these, 71 (18%) developed late GI symptoms and underwent endoscopy; these patients constituted the study population.
Demographics, disease, and transplant-related characteristics
are presented in Table 1. Among the GVHD-positive patients, 64% were males and 76% were Caucasian, as opposed to 50% and 96%, respectively, among those without confirmed gut GVHD. Additionally, 71% of GVHD-positive patients had received matched donor cells compared to 46% of GVHD-negative patients. The groups were similar on remaining variables.
Table 1.
Demographics and disease- and transplant-related characteristics
| Characteristic | Clinical Diagnosis | |
|---|---|---|
| GVHD (n=45) |
Not GVHD (n=26) |
|
| Age at transplant (years) (median, range) | 49, 21 – 70 | 48, 20 – 73 |
| Gender (male) (n, %) | 29 (64%) | 13 (50%) |
| Race | ||
| White (n, %) | 34 (76%) | 25 (96%) |
| Black (n, %) | 10 (22%) | 1 (4%) |
| Other (n, %) | 1 (2%) | 0 (0%) |
| Ethnicity (Hispanic) (n, %) | 2 (4%) | 0 (0%) |
| Primary disease | ||
| ALL, AML, MDS (n, %) | 24 (53%) | 14 (54%) |
| CML, CLL (n, %) | 5 (11%) | 4 (15%) |
| Lymphoma (n, %) | 11 (24%) | 5 (19%) |
| Other (n, %) | 5 (11%) | 3 (12%) |
| Donor | ||
| Matched-related (n, %) | 15 (33%) | 6 (23%) |
| Matched-unrelated (n, %) | 17 (38%) | 6 (23%) |
| Mismatched-related (haplo) (n, %) | 8 (18%) | 7 (27%) |
| Mismatched-unrelated (cord) (n, %) | 5 (11%) | 7 (27%) |
| Cell Source | ||
| Peripheral Blood Stem Cells (n, %) | 40 (89%) | 19 (73%) |
| Bone Marrow (n, %) | 0 (0%) | 0 (0%) |
| Umbilical Cord Blood (n, %) | 5 (11%) | 7 (27%) |
| Conditioning (Myeloablative) (n, %) | 7 (16%) | 3 (12%) |
| Prior allogeneic HSCT (yes) (n, %) | 3 (7%) | 2 (8%) |
| GVHD prophylaxis | ||
| MTX + CNI (n, %) | 24 (53%) | 12 (46%) |
| MMF-based (n, %) | 13 (29%) | 5 (19%) |
| Alemtuzamab-based (n, %) | 8 (18%) | 9 (35%) |
| On immunosuppression at time of endoscopy (yes) (n, %) | 40 (89%) | 22 (85%) |
| History of aGVHD (yes) (n, %) | 33 (73%) | 16 (62%) |
| History of aGHVD of the gut (yes) (n, %) | 14 (31%) | 7 (27%) |
| Median grade of aGVHD of the gut | 2 | 1 |
| History of cGVHD (yes) (n, %) | 6 (13%) | 3 (12%) |
| History of DLI (yes) (n, %) | 9 (20%) | 3 (12%) |
Abbreviations: GVHD: Graft Versus Host Disease; ALL: Acute Lymphocytic Leukemia; AML: Acute Myelogenous Leukemia; MDS: Myelodysplastic Syndrome; CML: Chronic Myelogenous Leukemia; CLL: Chronic Lymphocytic Leukemia; HSCT: Hematopoietic Stem Cell Transplant; MTX: Methotrexate; CNI: Calcineurin Inhibitor; MMF: Mycophenolate Mofteil; aGVHD: Acute Graft Versus Host Disease; cGVHD: Chronic Graft Versus Host Disease; DLI: Donor Lymphocyte infusion
Symptoms and endoscopy-related characteristics
are shown in Table 2. Patients with GVHD had a higher incidence of anorexia (53% versus 38%), abdominal pain/cramping (60% versus 42%), and diarrhea (87 % versus 54%) at presentation than those without GVHD. Compared to those without GVHD, a larger percent of those with GVHD had received Donor Lymphocyte Infusion (DLI) prior to endoscopy (20% versus 12%) and lost more weight since transplant (median of 13 versus 8 kg). Patients in the GVHD group received a median of 2 DLIs vs. 1 DLI for the non-GVHD group; the median dose was 1e7 CD3+ cells in both groups. The time to endoscopy was similar in both groups (160 days after transplant for those with GVHD, 188 days for those without).
Table 2.
Signs and symptoms at endoscopy
| Clinical Diagnosis | ||
|---|---|---|
| GVHD (n=45) |
Not GVHD (n=26) |
|
| Days from transplant to endoscopy (median, range) | 160 (100 – 917) | 188 (100 – 560) |
| If DLI, days from DLI to endoscopy (median, range) | 42, (15 – 133) | (14, 7 – 143) |
| Symptoms | ||
| Anorexia (n, %) | 24 (53%) | 10 (38%) |
| Abdominal pain or cramping (n, %) | 27 (60%) | 11 (42%) |
| Nausea or vomiting (n, %) | 28 (62%) | 16 (62%) |
| Diarrhea (n, %) | 39 (87%) | 14 (54%) |
| Weight loss (kg) (median, range) | −13.7, −100.8 to −1.0 | −8.3, −85.2 to −0.3 |
| Laboratory Findings | ||
| Platelet count at endoscopy(1 × 109 cells/L) (median, range) | 118, 16 – 396 | 98, 14 – 290 |
| Bilirubin at endoscopy(mg/dL) (median, range) | 0.70, 0.4 – 2.8 | 0.6, 0.2 – 1.6 |
| Albumin change from transplant to endoscopy (g/dL) (median, range) |
−0.4, −5.2 to 1.3 |
−0.2, −4.9 to +4.7 |
| Type endoscopy | ||
| EGD alone | 15 (33%) | 8 (31%) |
| Flex Sig alone | 11 (24%) | 10 (38%) |
| Colonoscopy alone | 8 (18%) | 1 (4%) |
| EGD + Flex Sig | 7 (16%) | 4 (15%) |
| EGD + Colonoscopy | 4 (9%) | 3 (12%) |
Abbreviations: GVHD: Graft Versus Host Disease; DLI: Donor Lymphocyte infusion
Histologic features
on biopsy are shown by group in Table 3. GVHD-positive patients had a significantly higher incidence of moderate to severe apoptosis (p=0.001), moderate to severe crypt destruction (p=0.02), and evidence of gland/crypt drop out (p=0.005) compared with GVHD-negative patients.
Table 3.
Histologic Findings
| Findings | Clinical Diagnosis | |
|---|---|---|
| GVHD (n=45) | Not GVHD (n=26) | |
| N (%) | N (%) | |
| Apoptosis (≥ grade 1) | 39 (87%) | 19 (73%) |
| Moderate to severe apoptosis (≥ grade 3) | 18 (40%) | 1 (4%) |
| Crypt destruction (≥ grade 1) | 25 (56%) | 5 (19%) |
| Moderate to severe crypt destruction (≥ grade 3) | 12 (27%) | 1 (4%) |
| Gland/crypt drop out | 15 (33%) | 1 (4%) |
| Gland/crypt atrophy | 11 (24%) | 2 (8%) |
| Lamina propria fibrosis/hyalinization | 8 (17%) | 2 (8%) |
| Collagen basement membrane deposition | 2 (4%) | 2 (8%) |
| Lamina propria inflammation (≥ grade 1) | 25 (56%) | 13(50%) |
| Cryptitis (≥ grade 1) | 15 (33%) | 8 (31%) |
| Ulcer | 6 (13%) | 2 (8%) |
| Intraepithelial lymphocytosis | 2 (4%) | 1 (4%) |
| Basal lymphoplasmacytosis | 0 (0%) | 0 (0%) |
| Lymphoid aggregates with associated injury | 2 (4%) | 3 (12%) |
| Lymphoid aggregates without associated injury | 6 (13%) | 4 (15%) |
| Crypt cell distortion | 7 (16%) | 1 (4%) |
| Paneth cell metaplasia | 4 (9%) | 1 (4%) |
| Loss of Paneth cells | 3 (7%) | 1 (4%) |
| Granulomata | 2 (4%) | 1 (4%) |
| Fibrosis | 1 (2%) | 0 (0%) |
| Vasculitis | 0 (0%) | 0 (0%) |
| Cytomegalovirus | 0 (0%) | 3 (12%) |
| Median, Range | Median, Range | |
| Number of eosinophils in the lamina propria | 17, 0 – 115 | 7, 0 – 118 |
| Number of eosinophils in the epithelium) | 0, 0 – 13 | 0, 0 – 19 |
Abbreviations: GVHD: Graft Versus Host Disease
Cause of symptoms
was GI-GVHD in 45 patients (63%), including late acute (n=39, 87%), classic chronic (n=1, 2%), and overlap (n=5, 11%). Of those that had GI symptoms but did not have GVHD, symptoms were attributed to infections (n=8, 11%), esophagitis, gastritis, or duodenitis (n=7, 10%), drugs (n=3, 4%), food intolerance (n=3, 4%), disease relapse (n=2, 3%), and delayed motility (n=1, 1%); in 2 (3%) cases, the etiology was unknown due to ambiguous findings on endoscopy and patient loss to follow-up. Of the 8 cases due to infections, 6 were due to cytomegalovirus (all positive by immunohistochemistry on biopsy), 1 due to Clostridium difficile (positive by polymerase chain reaction), and 1 consistent with infection on pathology but no organism was ever identified; however, the patient responded to antibiotics. For one of the CMV cases, the patient also had mild GI-GVHD, but GI symptoms improved with foscarnet treatment while immunosuppression was not increased, suggesting that CMV was the primary cause of GI symptoms. Of the 3 cases due to drugs, 1 was due to mycophenolate mofetil (MMF), 1 to ganciclovir, and 1 to ciprofloxacin and/or valganciclovir (both drugs were stopped with symptom improvement).
Variables related to GVHD
Of demographics and disease/transplant-related variables (category 1), those most highly related to GVHD were being white (p=0.055) and having received matched-donor cells (p=0.095). Among presenting signs and symptoms (category 2), the odds of having a diagnosis of GVHD were significantly higher for patients who presented with diarrhea compared to those who did not (p=0.0098) and having a higher pre-transplant bilirubin (p=0.065). Finally, the histologic feature (category 3) most predictive of GVHD was apoptosis. Each increase in apoptosis score was associated with increased risk of GVHD (p=0.0099, Table 4).
Table 4.
Predicting GVHD: Results of multivariate logistic regressions
| Model/Category of variables |
Variable | More: less likely to be GVHD+ |
OR (95% CI) |
p- value |
|---|---|---|---|---|
| 1= Demographics, disease/transplant | ||||
| Race | Non-White: White | 0.055 | ||
| Donor | Match: mismatch | 0.095 | ||
| Bilirubin | Higher: lower | 0.065 | ||
| 2= Signs and symptoms | ||||
| Diarrhea | Any: none | 0.0098 | ||
| 3= Pathologic features | ||||
| Apoptosis | Higher: lower score | 0.0099 | ||
| Final model | ||||
| Apoptosis | Higher: lower score | 2.35 (1.18 – 4.70) |
0.015 | |
| Diarrhea | Any: none | 5.43 (1.25 – 23.54) |
0.024 | |
| Donor | Match: mismatch | 3.49 (0.92 – 13.30) |
0.067 | |
Abbreviations: GVHD: Graft Versus Host Disease; CI: Confidence Interval
In the final multivariate analysis, apoptosis and diarrhea were most highly related to GVHD. Specifically, for each unit increase (worsening) in apoptosis score, the odds of a GVHD diagnosis more than doubled (OR=2.35, 95% CI=1.18 – 4.70, p=0.015). Patients who presented with diarrhea had a risk of GVHD that was more than five times that of patients without the symptom (OR=5.43 (1.25 – 23.54), p=0.024).
Overall survival and non-relapse mortality
At a median and maximum follow-up of 3.8 and 6.7 years, respectively, 44 (62%) patients had died, 12 from disease relapse and 32 from non-disease-related causes. Among all patients the median survival from first endoscopy was 8.5 months.
Figure 1a shows OS by diagnosis of GVHD. Median survival was three times longer for patients without GVHD compared to those with GVHD (22 versus 7 months). This difference was largely due to the higher incidence of NRM in GVHD patients compared to non-GVHD patients (p=0.079) (Figure 1b). For instance, at 6 months, the cumulative incidence of NRM was 31% for GVHD-positive patients and 19% for negative patients. By 12 months, these had increased to 49% versus 27%, respectively. Both groups had a similar cumulative incidence of relapse-related death (NS). Looking at the subgroup of patients with a history of acute GI GVHD, the cumulative incidence of NRM at 6 months was 50% for those with late GI-GVHD and 43% for those without late GI GVHD.
Figure 1a.
Overall survival by clinical diagnosis of GVHD
Figure 1b.
Cumulative incidence of non-relapse mortality by clinical diagnosis of GVHD
Results of multivariate analyses on OS showed that the variables that were significantly associated with OS were crypt dropout (p=0.012) and increased LP eosinophils (p=0.012) (Table 5). The hazard of dying was nearly two and a half times larger for patients whose biopsies showed drop-out compared to those without (HR=2.46, 95% CI=1.22 – 4.96). Each 5-unit decrease in eosinophils was associated with a 6% increase in risk of dying (HR=1.06, 95% CI=1.02 – 1.11). Degree of apoptosis was not related to increased mortality.
Table 5.
Predicting OS: Results of multivariate proportional hazards
| Variable | Comparison | HR (95% CI) | p-value |
|---|---|---|---|
| Clinical GVHD | Positive: negative | 1.64 (0.79 – 3.43) | 0.19 |
| Gland/crypt drop out | Yes: no | 2.46 (1.22 – 4.96) | 0.012 |
| LP eosinophils | Low: high | 1.06 (1.02 – 1.11)* | 0.012 |
For every 5 unit decrease
Abbreviations: OS: Overall Survival; CI: Confidence Interval; GVHD: Graft Versus Host Disease; LP: lamina propria
Corticosteroids
All 45 patients who were clinically positive for GVHD received steroid treatment. Nearly half these patients (22, 49%) had been on steroids prior to endoscopy. The median duration of steroids was 6 months, ranging from 17 days to 3.5 years. Thirty-one patients had died, 25 (81%) of whom were on steroids at time of death. Of these, the median duration of steroids was 4 months compared to 15 months for surviving patients.
Total Parenteral Nutrition (TPN)
A larger proportion of GVHD-positive patients received TPN compared to GVHD-negative patients (31% versus 8%, p=0.037). Among GVHD-positive patients, the presence of crypt destruction (p=0.0034) and larger weight loss (p=0.086) were most predictive of receiving TPN. The odds of receiving TPN for patients whose biopsies showed crypt destruction were two and a half times larger than those having biopsies without destruction (OR=2.56, 95% CI=1.37 – 4.81). For every 5 kgs of weight loss the odds of receiving TPN increased by 20% (OR=1.20, 95% CI=0.98 – 1.49). The two GVHD-negative patients who needed TPN had CMV gastritis and pancreatitis.
Mycophenolate Mofetil (MMF)
Of the total 71 patients, 34 (48%) received MMF prior to endoscopy; of these 34, 23 (68%) were diagnosed as GVHD-positive and the remaining 11 (32%) as GVHD-negative. Fifty-one percent of GVHD-positive patients compared to 42% of GVHD-negative patients received MMF (p=0.47). Histologically, the presence of moderate/severe apoptosis, moderate/severe crypt destruction, and gland/crypt drop out were correlated with a diagnosis of GVHD but did not correlate to MMF status (Table 6). Elevated LP eosinophils were seen in patients with GI-GVHD regardless of MMF administration, and LP eosinophil counts were not statistically different between MMF-negative and MMF-positive cases (p=0.84).
Table 6.
Histopathologic features of GI biopsies differentiated by MMF status
| Not GVHD | GVHD | |||
|---|---|---|---|---|
| −MMF (n=15) | +MMF (n=11) | −MMF (n=22) | +MMF (n=23) | |
| Moderate/severe apoptosis | 0 (0%) | 1 (9%) | 11 (50%) | 7 (30%) |
| Moderate/severe gland/crypt destruction | 0 (0%) | 1 (9%) | 7 (32%) | 5 (22%) |
| Gland/crypt drop out | 0 (0%) | 1 (9%) | 8 (36%) | 7 (30%) |
| LP eosinophils/HPF (median (range)) | 2 (0 – 11) | 1 (0 – 24) | 3 (0 – 23) | 3 (0 – 23) |
Abbreviations: MMF: Mycophenolate Mofetil; GVHD: Graft Versus Host Disease; LP: lamina propria; HPF: High Power Field
Discussion
Our study is the largest to date to look at late GI complications in patients with allogenic HSCT. While most studies on GI toxicities after HSCT focus on acute events within the first 100 days (13, 14) (with an emphasis on acute GVHD) or chronic GI GVHD, an advantage of this study is the focus on all late GI toxicities, both GVHD and non-GVHD, which are under-reported.
By 100 days after allogeneic transplant, most patients have recovered from conditioning and are often discharged home. Therefore, it was surprising to find that even among this “successful” group, 18% had late GI complications severe and/or persistent enough to warrant endoscopy. While most of these patients had GI-GVHD, more than a third had other causes explaining their symptoms. Therefore, while in the acute setting a high pre-test probability may lead to empiric steroids first and biopsy after, in the late setting investigation including endoscopy become more important in separating late GI-GVHD from other etiologies of GI distress.
It remains difficult to distinguish GI-GVHD from other significant GI complications based on clinical presentation alone. While the GVHD positive group had increased incidences of anorexia, abdominal pain, diarrhea, and weight loss when compared to the non-GVHD group, these symptoms were non-specific. While MMF is a known potential confounder of GI toxicity (15), we found no difference in MMF use in those with or without GVHD, nor did we find a difference in pathologic changes in the MMF positive versus MMF negative groups. As expected, apoptosis and gland/crypt drop out and destruction were among the most noted pathological findings of GVHD and contributed to the diagnosis of GVHD;however, these findings may still be present in patients without GVHD, especially when at minimal or mild levels. Therefore, attention needs to be paid to the whole constellation of findings: presentation, endoscopic findings, histology, and response to therapy and clinical outcome.
Late GI-GVHD was associated with significant non-relapse mortality (31% at 6 months and 49% at 12 months after endoscopy). Looking at steroid usage as a proxy for severity, 25 of 45 (55%) patients with GI-GVHD were still on steroids at death, and the median duration of steroid use for surviving patients was 15 months, emphasizing the persistent and refractory nature of this disease. Furthermore, 31% of these patients required TPN, compared to only 8% in the GVHD negative group. This was not surprising, for malnutrition is common in patients with GI-GVHD. In a study by Bassim et al (16), in patients with chronic GVHD, of whom 44% had GI-GVHD, the incidence of malnutrition was 29%. In our study patients with GI-GVHD also had significant drops in albumin (median 0.4 g/dL) and weight (median 13.7 kgs from start of transplant).
While non-relapse mortality was lower in patients without GVHD, it was still 19% at 6 months after endoscopy. This is even more noteworthy when considering that these patients were a median of 6 months after transplant at the time of their endoscopy and thus had survived the acute toxicities associated with pre-transplant conditioning. Infections were implicated in half of those deaths, suggesting problems with delayed immune reconstitution (17). At the same time, while infection with Bradyrhizobium enterica has been associated with cord colitis (18), and cord colitis has been reported in as many as 10% of cord blood transplants (19), we did not see any histopathologic cases of cord colitis among the 12 patients who received umbilical cord blood transplants and had late GI complications.
A limitation of this study is the retrospective design. Because we focused on late complications after 100 days, by which point most patients return to the care of their local doctor, we may have missed cases that were managed locally and not referred back to our institution, or cases that were treated empirically with resolution. However, this would only underestimate the prevalence of late GI complications. Another limitation is that it could be argued that day 100 is no longer a relevant breakpoint, and indeed most cases of GI-GVHD that we found were late acute GVHD or overlap. This emphasizes the importance of close follow up of patients after discharge to their local physician as late toxicities, GI or otherwise, frequently occur even when patients are more than 6 months out from transplant.
In conclusion, late GI complications are common following allogeneic stem cell transplant, even beyond the 100 day point when most patients are discharged from transplant. Physicians should be alert for both GVHD and non-GVHD etiologies as they can be difficult to distinguish clinically. For those who develop late GI-GVHD, the disease is often severe with frequent need for TPN, high non-relapse mortality, and prolonged duration of steroid therapy even for those who survive. Even non-GVHD GI complications are associated with significant non-relapse mortality.
Acknowledgements:
This study was supported by NCI 2P01-CA047741–16. Additional support was provided by the American Society of Hematology Research Training Award for Fellows, NHLBI 5T32-HL007057–37, and NCATS 5KL2TR001115–03 to ADS.
This study was approved by the Duke University Health Systems Institutional Review Board.
References
- 1.Wingard JR, Gastineau DA, Leather HL, Snyder E, Szczepiorkowski ZM (eds.). Hematopoietic Stem Cell Transplantation: A Handbook for Clinicians, 2nd Edition: AABB; 2015. [Google Scholar]
- 2.Barker CC, Anderson RA, Sauve RS, Butzner JD. GI complications in pediatric patients post-BMT. Bone marrow transplantation. 2005;36(1):51–8. [DOI] [PubMed] [Google Scholar]
- 3.Iqbal N, Salzman D, Lazenby AJ, Wilcox CM. Diagnosis of gastrointestinal graft-versus-host disease. The American journal of gastroenterology. 2000;95(11):3034–8. [DOI] [PubMed] [Google Scholar]
- 4.Appelbaum FR, Forman SJ, Negrin RS, Antin JH. Thomas’ hematopoietic cell transplantation : stem cell transplantation. Fifth edition. ed. p. p.
- 5.Schulenburg A, Turetschek K, Wrba F, Vogelsang H, Greinix HT, Keil F, et al. Early and late gastrointestinal complications after myeloablative and nonmyeloablative allogeneic stem cell transplantation. Annals of hematology. 2004;83(2):101–6. [DOI] [PubMed] [Google Scholar]
- 6.Akpek G, Chinratanalab W, Lee LA, Torbenson M, Hallick JP, Anders V, et al. Gastrointestinal involvement in chronic graft-versus-host disease: a clinicopathologic study. Biology of blood and marrow transplantation : journal of the American Society for Blood and Marrow Transplantation. 2003;9(1):46–51. [DOI] [PubMed] [Google Scholar]
- 7.Sung AD, Chao NJ. Concise review: acute graft-versus-host disease: immunobiology, prevention, and treatment. Stem cells translational medicine. 2013;2(1):25–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gooley TA, Chien JW, Pergam SA, Hingorani S, Sorror ML, Boeckh M, et al. Reduced mortality after allogeneic hematopoietic-cell transplantation. The New England journal of medicine. 2010;363(22):2091–101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Wild D, Sung AD, Cardona D, Cirricione C, Sullivan K, Detweiler C, et al. The Diagnostic Yield of Site and Symptom-Based Biopsies for Acute Gastrointestinal Graft-Versus-Host Disease: A 5-Year Retrospective Review. Digestive diseases and sciences. 2016;61(3):806–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Rowlings PA, Przepiorka D, Klein JP, Gale RP, Passweg JR, Henslee-Downey PJ, et al. IBMTR Severity Index for grading acute graft-versus-host disease: retrospective comparison with Glucksberg grade. British journal of haematology. 1997;97(4):855–64. [DOI] [PubMed] [Google Scholar]
- 11.Shulman HM, Cardona DM, Greenson JK, Hingorani S, Horn T, Huber E, et al. NIH Consensus development project on criteria for clinical trials in chronic graft-versus-host disease: II. The 2014 Pathology Working Group Report. Biology of blood and marrow transplantation : journal of the American Society for Blood and Marrow Transplantation. 2015;21(4):589–603. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Shulman HM, Kleiner D, Lee SJ, Morton T, Pavletic SZ, Farmer E, et al. Histopathologic diagnosis of chronic graft-versus-host disease: National Institutes of Health Consensus Development Project on Criteria for Clinical Trials in Chronic Graft-versus-Host Disease: II. Pathology Working Group Report. Biology of blood and marrow transplantation : journal of the American Society for Blood and Marrow Transplantation. 2006;12(1):31–47. [DOI] [PubMed] [Google Scholar]
- 13.Nomura K, Iizuka T, Kaji D, Yamamoto H, Kuribayashi Y, Kimura R, et al. Clinicopathological features of patients with acute graft-versus-host disease of the upper digestive tract. Journal of gastroenterology and hepatology. 2014;29(11):1867–72. [DOI] [PubMed] [Google Scholar]
- 14.Ross WA, Ghosh S, Dekovich AA, Liu S, Ayers GD, Cleary KR, et al. Endoscopic biopsy diagnosis of acute gastrointestinal graft-versus-host disease: rectosigmoid biopsies are more sensitive than upper gastrointestinal biopsies. The American journal of gastroenterology. 2008;103(4):982–9. [DOI] [PubMed] [Google Scholar]
- 15.Iida M, Fukuda T, Ikegame K, Yoshihara S, Ogawa H, Taniguchi S, et al. Use of mycophenolate mofetil in patients received allogeneic hematopoietic stem cell transplantation in Japan. International journal of hematology. 2011;93(4):523–31. [DOI] [PubMed] [Google Scholar]
- 16.Bassim CW, Fassil H, Dobbin M, Steinberg SM, Baird K, Cole K, et al. Malnutrition in patients with chronic GVHD. Bone marrow transplantation. 2014;49(10):1300–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Kanda J, Chiou LW, Szabolcs P, Sempowski GD, Rizzieri DA, Long GD, et al. Immune recovery in adult patients after myeloablative dual umbilical cord blood, matched sibling, and matched unrelated donor hematopoietic cell transplantation. Biology of blood and marrow transplantation : journal of the American Society for Blood and Marrow Transplantation. 2012;18(11):1664–76 e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bhatt AS, Freeman SS, Herrera AF, Pedamallu CS, Gevers D, Duke F, et al. Sequence-based discovery of Bradyrhizobium enterica in cord colitis syndrome. The New England journal of medicine. 2013;369(6):517–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Herrera AF, Soriano G, Bellizzi AM, Hornick JL, Ho VT, Ballen KK, et al. Cord colitis syndrome in cord-blood stem-cell transplantation. N Engl J Med. 2011;365(9):815–24. [DOI] [PubMed] [Google Scholar]