Clinical impact of endoscopy in severely thrombocytopenic patients with hematologic malignancy experiencing gastrointestinal bleeding

Abstract

BACKGROUND

Gastrointestinal bleeding (GIB) is a major cause of hospitalization worldwide. Patients with hematologic malignancies have a higher risk of GIB as a result of thrombocytopenia and platelet dysfunction. There is no consensus on the optimal platelet level that would be safe for endoscopic intervention, although a platelet level of > 50 × 10⁹ / L was suggested based on expert opinion. There is a paucity of data on whether endoscopic intervention and the timing of endoscopy impacted the outcome of patients with hematologic malignancy and severe thrombocytopenia who experienced acute overt GIB.

AIM

To assess the safety of endoscopic intervention of inpatients with hematological malignancies and severe thrombocytopenia presenting with acute overt GIB.

METHODS

This is a single center retrospective study. The data was collected from the electronic health record from 2018 to 2020. Inpatients with hematologic malignancy who presented with acute overt GIB and platelet count ≤ 50 × 10⁹/L were included in the study. Outcomes included mortality, transfusion requirements, length of stay, intensive care unit admission and recurrent bleeding. A subgroup analysis was performed to compare the outcomes of urgent endoscopy within 24 hours of GIB vs endoscopy > 24 hours.

RESULTS

A total of 76 patients were identified. The mean platelet count is 24.3 in the endoscopy arm and 14.6 in the conservative management arm. There was no statistically significant difference between patients who had endoscopy vs conservative management in 30-day (P = 0.13) or 1 year (P = 0.78) mortality, recurrent bleeding (P = 0.68), transfusion of red blood cells (P = 0.47), platelets (P = 0.31), or length of stay (P = 0.94). A subgroup analysis comparing urgent endoscopy within 24 hours compared with delayed endoscopy showed urgent endoscopy was not associated with improved 30-day or 1 year mortality (P = 0.11 and 0.46, respectively) compared to routine endoscopy, but was associated with decreased recurrent bleeding in 30 days (P = 0.01).

CONCLUSION

Medical supportive treatment without endoscopy could be considered as an alternative to endoscopic therapy for patients with hematologic malignancy complicated by severe thrombocytopenia and acute non-variceal GIB.

Key Words: Endoscopy; Gastrointestinal bleeding; Gastrointestinal bleeding mortality; Hematologic malignancy; Thrombocytopenia

Core Tip: Hematologic malignancies have a higher risk of diffuse gastrointestinal bleeding (GIB) related to severe thrombocytopenia and platelet dysfunction. It is unclear if endoscopy impacts the management of these patients. In this study, we investigated whether endoscopy improved the outcome of patients with hematologic malignancy and severe thrombocytopenia who experienced acute overt GIB compared to conservative management. Our results show no statistical difference in mortality, recurrent bleeding and length of stay between endoscopic and conservative management. However, endoscopy within 24 hours of GIB significantly reduced the 30-day recurrent bleeding rate compared to endoscopy performed > 24 hours.

INTRODUCTION

Gastrointestinal bleeding (GIB) is a major cause of hospitalization worldwide. In the United States, GIB accounts for approximately 224 cases per 100000 admissions with a mortality rate of 2.7%[1]. Patients with hematologic malignancies are at heightened risk of GIB due to thrombocytopenia and platelet dysfunction, which may result from the malignancy itself or cytotoxic agents used in the therapy[2]. There is no consensus on the optimal platelet level that would be safe for endoscopic intervention, although a platelet level of > 50 × 10⁹/L was suggested based on expert opinion[3]. Clinicians are often faced with a patient with ongoing GIB requiring repeated transfusions, borderline hemodynamic stability, and a need for further evaluation and treatment by endoscopy. In the case of hematologic malignancy with severe thrombocytopenia, platelet levels can be limited by platelet refractoriness, bone marrow infiltration, malnutrition and malignancy-induced autoantibodies[4]. When endoscopic therapy is performed after transfusions, platelet levels can relapse into the severe thrombocytopenia range and the patient may have recurrent GIB.

The role of endoscopy in this context is debated. One school of thought is that bleeding may be primarily driven by thrombocytopenia, which may be better treated with transfusion and avert the need for endoscopic evaluation and its risks. Conversely, another view is that severe thrombocytopenia unmasks coexistent and intervenable gastrointestinal pathology. Prior studies in solid and hematologic cancer patients with acute GIB and thrombocytopenia have shown that endoscopy identified unifocal bleeding sources in more than 50% of cases, but diffuse bleeding sources were more common in patients with severe thrombocytopenia[5]. In one of the largest studies on endoscopy in cancer patients with severe thrombocytopenia, active GIB was found in only 11% of patients; however, subgroup analysis of the patients with hematologic malignancy was not presented[6]. To our knowledge, examination of this challenging scenario focusing on the benefit of endoscopy for patients with hematologic malignancies complicated by severe thrombocytopenia and acute GIB has not been explored.

Current guidelines recommend upper endoscopic intervention within 24 hours for upper GIB. Additionally, for lower GIB with severe hematochezia and negative initial upper endoscopy, colonoscopy after rapid bowel prep is recommended within 24 hours of admission[7,8]. However, the urgency of endoscopic intervention remains controversial with conflicting data[9,10]. Furthermore, the timing of endoscopy in non-variceal GIB for severely thrombocytopenic patients has not been well-defined. In this study, we examine the clinical outcome of endoscopic therapy compared to conservative management in patients with acute overt non-variceal GIB and severe thrombocytopenia secondary to hematologic malignancy. We also explore the impact of urgent vs non-urgent endoscopic intervention in this patient population.

MATERIALS AND METHODS

Study design

This is a single centered, retrospective study on whether endoscopic intervention reduces mortality compared to conservative management among admitted patients with acute overt non-variceal GIB and severe thrombocytopenia secondary to hematologic malignancy. The study was approved by the institutional review board before collecting and analyzing data. Further subgroup analysis comparing patients undergoing urgent endoscopic intervention within 24 hours of index GIB event to endoscopic intervention beyond 24 hours.

Study population and data collection

Our study population was identified as inpatients with an underlying hematologic malignancy and severe thrombocytopenia who developed an overt GIB. Hematological malignancy included leukemia and subtypes, lymphoma and subtypes, myelofibrosis, multiple myeloma, myelodysplastic syndrome and aplastic anemia. Severe thrombocytopenia was defined as platelet count < 50 × 10³/mL. Overt GIB was defined as hematemesis, melena, or hematochezia. Data were retrospectively collected from the electronic health record using the Epic database from the years 2018-2020. Each patient’s chart was reviewed to collect basic demographics, clinical information, and endoscopic findings. Exclusion criteria include age less than 18 years old, cirrhosis, outpatients, occult bleeding, solid non-hematologic tumor, missing information, no underlying hematologic disease, and patients with platelet count > 50 × 10³/mL. Urgent endoscopy was defined as endoscopy performed within 24 hours of the onset of GIB.

Outcome measures

The primary outcome is 30-day and one-year mortality after the index episode of GIB. Secondary outcomes included intensive care unit (ICU) admission secondary to GIB; packed red blood cell (RBC) transfusion; platelet transfusion; recurrent overt bleeding within 30 days of the GIB; and length of stay.

Statistical analysis

We provide descriptive statistics including means and standard deviations for continuous variables and percentages for categorical variables. We analyzed the association of each predictor with the outcome using independent t-test for continuous variables, and Fisher’s Exact test for categorical data. Statistical significance was presented with test statistics and P values. All analyses were done using SPSS version 29.

RESULTS

Baseline demographics and clinical characteristics

A total of 76 patients fulfilling study criteria were included in the study, with 38 patients (50%) undergoing endoscopy by chance. Demographic and baseline characteristics of patients who underwent endoscopic intervention vs patients who did not were shown in Table 1. There was no statistical difference in age, gender, body mass index (BMI), ethnicity, comorbidities, presentation of GIB, and hematological malignancies between the 2 groups. However, the mean hemoglobin was significantly higher (7.1 ± 1 g/dL vs 6.3 ± 1.6 g/dL, P = 0.01) and the mean platelet count was significantly lower (14.6 ± 13.3 vs 24.3 ± 13.7 platelets/mcL, P = 0.01) in the group of patients who did not have endoscopy, when compared to the group of patients who had endoscopy (Table 2). Other clinical and laboratory variables such as systolic blood pressure, diastolic blood pressure, heart rate, white blood count, international normalized ratio and partial thromboplastin time were comparable between the 2 groups with no statistical significance (Table 2).

Table 1 Baseline demographics and characteristics, mean ± SD or n (%).

Variable	No endoscopy, n = 38	Endoscopy, n = 38	P value
Age	62.7 ± 10.5	58.5 ± 15.6	0.17
Gender
Male	24 (63.2)	27 (71.1)	0.46
Female	14 (36.8)	11 (28.9)
BMI	30.8(12.7)	27.2(8.9)	0.15
Ethnicity
White	32 (84.2)	26 (68.4)	0.24
Black	5 (13.2)	11 (28.9)
Asian	1 (2.6)	1 (2.7)
Comorbidities
Hypertension	25 (65.8)	21 (55.3)	0.35
Cardiac	7 (18.4)	13 (34.2)	0.12
Pulmonary	8 (21.1)	3 (7.9)	0.1
CKD	8 (21.1)	7 (8.4)	0.77
Presentation of GI bleeding
Melena	22 (57.9)	30 (78.9)	0.08
Hematemesis	8(21.1)	2 (5.3)	0.09
Hematochezia	8(21.1)	6 (15.8)	0.77
Hematologic disease/malignancy
MDS	7 (18.4)	6 (15.8)	0.76
AML	11 (28.9)	12 (31.6)	0.8
CML	1 (2.6)	3 (7.9)	0.3
ALL	6 (15.8)	5 (13.2)	0.74
Multiple myeloma	7 (18.4)	7 (18.4)	1.0
NHL	2 (5.3)	2 (5.3)	1.0
CLL	1 (2.6)	0 (0)	0.31
Aplastic anemia	2 (5.3)	2 (5.3)	1.0
Myelofibrosis	1 (2.6)	1 (2.6)	1.0

BMI: Body mass index; CKD: Chronic kidney disease; GI: Gastrointestinal; MDS: Myelodysplastic syndrome; AML: Acute myeloid leukemia; CML: Chronic myeloid leukemia; ALL: Acute lymphoblastic leukemia; NHL: Non-Hodgkin lymphoma; CLL: Chronic lymphocytic leukemia.

Table 2 Basic clinical and laboratory characteristics of patients.

Variable	No Endoscopy, n = 38	Endoscopy, n = 38	P value
SBP	109.6 ± 19.7	116.3 ± 23.3	0.18
DBP	62.9 ± 13.3	66.6 ± 14.7	0.25
HR	105.2 ± 20.7	98.4 ± 17.2	0.12
HGB	7.1 ± 1	6.3 ± 1.6	0.01a
WBC	6.4 ± 16.6	6.2 ± 9.6	0.97
PLT	14.6 ± 13.3	24.3 ± 13.7	0.01a
INR	1.3 ± 0.3	1.5 ± 1.3	0.46
PTT	32 ± 12.4	34.4 ± 16.4	0.48

^aP < 0.05.

SBP: Systolic blood pressure; DBP: Diastolic blood pressure; HR: Heart rate; HGB: Hemoglobin; WBC: White blood count; PLT: Platelet; INR: International normalized ratio; PTT: Partial thromboplastin time.

Outcome

Our primary outcome was the 30-day and one-year mortality rate after the index GIB episode. There was no statistically significant difference between patients in the endoscopy group compared to the non-endoscopy group in the 30 day (36.8% vs 21%, P = 0.13) or one year mortality (73.7% vs 71.1%, P = 0.78; Table 3). There was also no significant difference between the endoscopy group and non-endoscopy group in the secondary outcomes, including admission to ICU secondary to GIB, requirement of RBC or platelet transfusion, recurrent GI bleeding, and length of stay (Table 3).

Table 3 Outcomes of patients treated with medical management vs endoscopy, mean ± SD or n (%).

Variable	No endoscopy, n = 38	Endoscopy, n = 38	P value
Mortality at 30 days	8 (21.0)	14 (36.8)	0.13
Mortality at 1 year	27 (71.1)	28 (73.7)	0.78
ICU admission	12 (31.6)	14 (36.8)	0.63
PRBC transfusion	31 (81.6)	34 (89.5)	0.47
PLT transfusion	38 (100)	37 (97.4)	0.31
Recurrent bleeding	13 (34.2)	14 (36.8)	0.68
Length of stay (day)	18.2 ± 14.9	17.8 ± 25.7	0.94

ICU: Intensive care unit; PRBC: Packed red blood cells; PLT: Platelet.

Subgroup-analysis of urgent vs non-urgent endoscopy

The patients who had endoscopies (n = 38) were further divided into patients undergoing endoscopy within 24 hours of GIB (n = 19) and patients undergoing endoscopy after 24 hours (n = 19). The average systolic blood pressure in the > 24-hour endoscopy group was higher (125.3 ± 21.6 mmHg) than the < 24-hour endoscopy group (107.4 ± 21.8 mmHg; P = 0.02; Table 4). Otherwise, there was no significant difference in the levels of hemoglobin and platelet between the 2 groups. Patients undergoing endoscopy > 24 hours had significantly higher recurrent bleeding within 30 days (57.9%) than patients undergoing endoscopy < 24 hours (15.8%; P = 0.01; Table 5). The 30-day and one-year mortality rate, ICU admission, transfusion requirement, and length of stay were all similar in both groups (Table 5). Table 6 outlined the endoscopic interventions performed and Table 7 listed the findings of endoscopy in patients who had endoscopy urgently vs non-urgently.

Table 4 Basic clinical and laboratory characteristics of patients treated with endoscopic intervention urgently vs non-urgently.

Variable	Endoscopy < 24 hours, n = 19	Endoscopy > 24 hours, n = 19	P value
SBP	107.4 ± 21.8	125.3 ± 21.6	0.02a
DBP	62.6 ± 15.4	70.6 ± 13.2	0.1
HR	104.8 ± 15.8	92 ± 16.4	0.02a
HGB	6.1 ± 1.4	6.5 ± 1.7	0.41
WBC	9.2 ± 12.8	3.3 ± 2.8	0.06
PLT	25.7 ± 14	22.8 ± 13.7	0.52
INR	1.7 ± 1.8	1.3 ± 0.3	0.32
PTT	36.4 ± 18.8	32.4 ± 13.7	0.46

^aP < 0.05.

SBP: Systolic blood pressure; DBP: Diastolic blood pressure; HR: Heart rate; HGB: Hemoglobin; WBC: White blood count; PLT: Platelet; INR: International normalized ratio; PTT: Partial thromboplastin time; SD: Standard deviation.

Table 5 Outcome of patients treated with endoscopy urgently vs non-urgently, mean ± SD or n (%).

	Endoscopy < 24 hours, n = 19	Endoscopy > 24 hours, n = 19	P value
Mortality at 30 days	6 (31.6)	2 (10.5)	0.11
Mortality at 1 year	4 (21.1)	6 (31.6)	0.46
ICU admission	9 (47)	5 (26.3)	0.18
PRBC transfusion	17 (89.5)	17 (89.5)	1.0
PLT transfusion	19 (100)	18 (94.7)	0.31
Recurrent bleeding	3 (15.8)	11 (57.9)	0.01a
Length of stay (days)	13.5 ± 13	22.1± 33.9	0.31

^aP < 0.05.

ICU: Intensive care unit; PRBC: Packed red blood cells; PLT: Platelet.

Table 6 Type of endoscopic intervention, n (%).

Variable	Endoscopy < 24 hours, n = 19	Endoscopy > 24 hours, n = 19	P value
No intervention	14 (73.7)	15 (78.9)	0.72
Intervention	5 (26.3)	4 (21.1)	0.7
Thermal	1	3
Mechanical	3	0
Dual	0	1
Banding	1	0

Table 7 Findings on endoscopy, n (%).

Variable	Endoscopy < 24 hours	Endoscopy > 24 hours
EGD findings	n = 16	n = 17
Normal	1 (6.3)	6 (35.3)
PUD	3 (18.8)	2 (11.8)
AVM	1 (6.3)	2 (11.8)
Mucosal inflammation	5 (31.3)	3 (17.6)
PHG	2 (12.5)	0
Solid tumor	1 (6.3)	0
Dieulafoy	2 (12.5)	0
Mallory Weiss	0	1 (5.9)
Cameron lesion	0	2 (11.8)
Esophagitis	1 (6.3)	1 (5.9)
Colonoscopy findings	n = 6	n = 5
Normal	2 (33.3)	2 (40)
Blood with no source	1 (16.7)	2 (40)
Radiation proctitis	0	1 (20)
Mucosal ulcer	1 (16.7)	0
Mucosal inflammation	2 (33.3)	0

PUD: Peptic ulcer disease; AVM: Arteriovenous malformation; PHG: Portal hypertensive gastropathy.

DISCUSSION

In this retrospective study, endoscopic intervention was not associated with improvement in mortality at 30 days or one year compared to medical management for patients with hematological malignancies presenting with acute overt non-variceal GIB and severe thrombocytopenia. This suggests that the primary driver of patient outcome in these patients could be the underlying hematological malignancies. GIB may merely reflect the severity of underlying disease and thus treating GIB does not impact the course and outcome. Our results suggest that patients with hematological malignancy and severe thrombocytopenia may not benefit under the current guideline recommendations for urgent endoscopy in upper GIB patients with a history of malignancy[7]. In addition, endoscopy and anesthesia introduce potential risks in this highly comorbid population. The risks and benefits should be carefully considered in this patient population before performing endoscopy.

The appropriateness of pre-procedural platelet targets > 50 × 10⁹/L for patients with hematologic malignancy remains unclear. Ramos et al[11] looked at a cohort of 144 patients evaluating the safety of endoscopic intervention in thrombocytopenic patients presenting with overt GIB. The median PLT count was 41 × 10⁹. The study showed increasing overall mortality and recurrent bleeding in the endoscopy group when compared to conservative management. However, their cohort included 88 cirrhotic patients whereas cirrhosis is an exclusion criterion in our study. Krishna et al[6] assessed transfusion requirement in a cohort of 68 patients with a primary diagnosis of hematologic malignancy who presented with GIB and thrombocytopenia and found a reduction in transfusion requirements in patients undergoing endoscopic intervention. The patients in this study had a primary hematologic malignancy but they included patients with platelet count of 75 × 10⁹ or lower, but the platelet cutoff in our study is 50 × 10⁹.

Patients with hematologic malignancy with severe thrombocytopenia are not only at higher risk for spontaneous or more severe GIB from structural abnormalities but are also introduced to further bleeding risk undergoing endoscopic procedures as well as reduced efficacy of any interventions deployed[12]. Further, severe thrombocytopenia in hospitalized patients with hematologic malignancy can often be multifactorial and with acutely reversible contributing factors such as sepsis, dilution, or medications (i.e., heparin). Higher hemoglobin levels and lower platelet counts were noted in the non-endoscopy group (Table 2). The higher perceived procedural risks associated with lower platelet count coupled with higher hemoglobin level may tip the risks and benefits calculation against endoscopy. Nevertheless, the lower platelet counts in the non-endoscopy group were not associated with higher mortality, which again supports the view that endoscopic treatment may have limited benefit in severely thrombocytopenic patients with hematologic malignancy. Continued supportive measures by transfusion, addressing reversible causes of thrombocytopenia and ongoing definitive management of the hematologic malignancy may be more beneficial to reduce the bleeding diathesis before an endoscopic approach is attempted. This approach is supported by our findings that endoscopic intervention did not impact mortality, transfusion need, recurrent GIB, or hospital stay.

Urgent endoscopy within 24 hours was not associated with decreased mortality, transfusion requirements or hospital stay. However, recurrent bleeding at 30 days was reduced. Since our data shows that endoscopic intervention did not improve patient outcomes compared to medical management, should endoscopy be determined to be performed, it appears that urgent endoscopy would provide more benefits with less recurrent bleeding.

The participants in this study reflected the broader patient population demographically with no significant differences in age, sex, or ethnicity. However, there was a strong trend toward more pulmonary comorbidities and higher BMI in the conservative management group, probably reflecting concern for procedural and anesthesia tolerance. The acuity of patients between the endoscopic and non-endoscopic groups were similar given no significant difference of ICU admission or hemodynamic profiles was observed.

Our study was limited by its retrospective single-center nature and modest sample size which may reduce our power to detect smaller differences. The hematologic malignancy included in the study was primarily composed of acute myeloid leukemia, multiple myeloma, and myelodysplastic syndrome, with less representation of other diseases such as chronic lymphocytic leukemia. However, this is consistent with the baseline differences in the condition’s needs for inpatient admission. Finally, as a single-center study in a large metropolitan area, our study may underestimate the recurrent bleeding rate because patients may receive care from other healthcare systems within the region, particularly for emergencies. However, mortality rates are generally preserved across systems due to a central state reporting process.

Future studies could include a prospective multi-center design as well as exploring the reproducibility of our findings across several thrombocytopenia thresholds. Additionally, incorporating platelet function into decision-making with a point of care assay such as thromboelastography may better delineate safe procedural hemostasis parameters.

CONCLUSION

We found that endoscopy was not associated with improved mortality in patients with acute overt non-variceal GIB with hematologic malignancy and severe thrombocytopenia in comparison with conservative management. Our findings suggest that medical supportive treatment without endoscopy could be considered as a reasonable alternative to endoscopic therapy for those with hematologic malignancy complicated by severe thrombocytopenia and GIB.