Lethal small bowel necrosis due to aspergillosis during acute promyelocytic leukemia induction


  • Laura E. Lunde,

    1. Division of Hematology–Oncology and Transplantation, University of Minnesota, Minneapolis, Minnesota
    2. Department of Medicine, University of Minnesota, Minneapolis, Minnesota
    Search for more papers by this author
  • Charles Chuang,

    1. Department of Medicine, University of Minnesota, Minneapolis, Minnesota
    Search for more papers by this author
  • Michael A. Linden,

    1. Divison of Hematopathology, Department of Laboratory Medicine and Pathology, University of Minnesota, Minneapolis, Minnesota
    Search for more papers by this author
  • Sarah A. Williams,

    1. Divison of Hematopathology, Department of Laboratory Medicine and Pathology, University of Minnesota, Minneapolis, Minnesota
    Search for more papers by this author
  • Zohar Sachs,

    1. Division of Hematology–Oncology and Transplantation, University of Minnesota, Minneapolis, Minnesota
    2. Department of Medicine, University of Minnesota, Minneapolis, Minnesota
    Search for more papers by this author
  • Zuzan Cayci,

    1. Department of Radiology, University of Minnesota, Minneapolis, Minnesota
    Search for more papers by this author
  • Jo-Anne H. Young,

    1. Department of Medicine, University of Minnesota, Minneapolis, Minnesota
    2. Division of Infectious Disease, Department of Medicine, University of Minnesota, Minneapolis, Minnesota
    Search for more papers by this author
  • Celalettin Ustun

    Corresponding author
    1. Division of Hematology–Oncology and Transplantation, University of Minnesota, Minneapolis, Minnesota
    2. Department of Medicine, University of Minnesota, Minneapolis, Minnesota
    • Correspondence to: Celalettin Ustun, MD, Associate Professor of Medicine, Division of Hematology Oncology and Transplantation, Department of Medicine, University of Minnesota, 14-142 PWB, 516 Delaware Street SE, Minneapolis, MN 55455. E-mail: custun@umn.edu

    Search for more papers by this author

  • Conflict of interest: Nothing to report

  • A physician or group of physicians considers presentation and evolution of a real clinical case, reacting to clinical information and data (boldface type). This is followed by a discussion/commentary

A 45-year-old female presented with persistent headache and generalized bruising. A complete blood count (CBC) showed mild anemia (11.5 g dL−1), thrombocytopenia (15 × 109/L), leukocytosis (43 × 109/L) with promyelocyte predominance (38.3 × 109/L), and a coagulopathy with INR 1.93, D-dimer > 20 μg mL−1 fibrinogen-equivalent units (FEU) (0.0–0.50 μg mL−1 FEU), and fibrinogen 150 mg dL−1 (200–420 mg dL−1). Peripheral blood and marrow aspirate smears demonstrated atypical promyelocytes with hypogranulation and multiple Auer rods (Fig. 1A). Myeloperoxidase cytochemical stain performed on the peripheral blood showed the atypical cells to be of myeloid origin (Fig. 1B). Bone marrow biopsy and aspiration showed cellularity of nearly 100% that was diffusely and extensively effaced by leukemic promyelocytes (Fig. 1C,D). Flow cytometry showed blasts expressing CD13, CD33, partial CD34, CD45, partial CD117, and partial cytoplasmic myeloperoxidase, consistent with a clonal myeloid neoplasm. Florescent in situ hybridization (FISH) analysis indicated that 95.5% of interphase cells demonstrated promyelocyte leukemia/retinoic acid receptor alpha (PML-RARA) fusion gene. G-banding karyotyping was reported as 46,XX,t(15;17)(q24.1;q21)[18]/46,XX[2].

Figure 1.

Blood and bone marrow samples from patient at diagnosis. A: Diagnostic peripheral smear with leukemic promyelocytes with multiple Auer rods (left) and bilobed hypogranulated forms (Wright-Giemsa ×100 oil). B: Leukemic promyelocytes (right) staining positive for myeloperoxidase (MPO) cytochemical stain along with a non-neoplastic neutrophil staining with MPO and serving as an internal control (Wright-Giemsa ×50 oil). C: Bone marrow concentrated aspirate with clusters of numerous leukemic hypolobated promyelocytes with bilobed nuclei and hypogranulated cytoplasm (Wright-Giemsa ×100 oil). D: Bone marrow trephine biopsy showing diffuse replacement by leukemic promyelocytes (hematoxylin and eosin ×10). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

The patient had high risk acute promyelocytic leukemia (APL), which is characterized by the morphologic predominance of leukemic promyelocytes and a t(15;17)(q22;q12) involving the retinoic acid receptor alpha gene. High-risk APL is characterized by a white blood cell count (WBC) at presentation of >10 × 109/L and is associated with a high risk of death during induction therapy and relapse. Vitamin A derivative all-trans retinoic acid (ATRA) is mainstay of APL treatment and APL today is considered to be the most curable subtype of all AMLs.

The patient was started on all-trans retinoic acid (ATRA) 45 mg m−2, idarubicin 12 mg m−2 Days 1–4, and dexamethasone 10 mg IV every 6 hr for high risk acute promyelocytic leukemia (APL). Supportive care was provided for the patient's coagulopathy with fresh frozen plasma, cryoprecipitate and platelet transfusions. Prophylactic antimicrobial therapy was initiated with acyclovir, levofloxacin, and fluconazole. On Day 6 of induction chemotherapy (D+6), the patient was noted to have developed peripheral edema and significant weight gain. Baseline radiograph of the chest demonstrated pleural effusions.

These findings were thought due to mild ATRA or differentiation syndrome, which is generally characterized by fever and respiratory distress. However, it is also associated with weight gain, lower extremity edema, pleural or pericardial effusions, and episodic hypotension. The patient continued treatment with dexamethasone.

On D+10, she developed neutropenic fever, with isolation of methicillin-resistant Staphylococcus aureus (MRSA) in blood cultures. Intravenous vancomycin and ceftazidime were started with resolution of her fever. The patient again became febrile on D+19. Chest-computed tomography (CT) showed bilateral small pleural effusions in the setting of a steroid taper. ATRA was discontinued due to concern for worsening ATRA syndrome and steroids were resumed. On this day, the patient developed abdominal distention and emesis. Abdominal CT demonstrated small bowel obstruction, thought to be secondary to ischemia, which was managed conservatively in the setting of her ongoing neutropenia and clinical stability.

The patient's antibiotic regimen was adjusted to metronidazole and piperacillin/tazobactam. Intravenous voriconazole replaced fluconazole on D+26. Additionally, she was started on granulocyte colony-stimulating factor. During this time, she noted double vision and color distortion. A head CT was unremarkable and brain magnetic resonance imaging (MRI) was delayed secondary to increased respiratory distress, which the patient developed in the setting of continued abdominal distention. She was intubated and repeat abdominal CT imaging showed a proximal closed loop small bowel obstruction (Fig. 2).

Figure 2.

Coronal reconstruction image of a contrast enhanced abdomen-pelvis CT demonstrates small bowel obstruction and ascites (*). The arrow points to the transition point. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Exploratory laparotomy was performed, with resection of a portion of proximal small bowel. At the time of surgery, the proximal jejunum was grossly necrotic. Small bowel pathology on D+32 showed transmural necrosis, with numerous invasive septated fungal organisms consistent with Aspergillus species and extensive angioinvasion by these organisms (Fig. 3A–D). Aspergillus galactomannan testing of serum was negative on D+32.

Figure 3.

Histopathological analysis of bowel resection specimen. A–D: Duodenal resection specimen. A: Extensive necrosis, ulceration with residual mucosa (hematoxylin and eosin ×10). B: Angioinvasion by fungal forms (hematoxylin and eosin ×40). C: Extensive submucosal neutrophilic infiltration with fungal forms (hematoxylin and eosin ×40). D: Fungal forms with septations (arrow) (hematoxylin and eosin ×100 oil). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Bone marrow biopsy on D+33 showed that she was in hematologic CR, and FISH was negative for PML-RARA fusion gene. ATRA was restarted. Possible seizure activity was noted, and a lumbar puncture (LP) and brain MRI were performed. MRI showed subarachnoid hemorrhage in the right occipital and parietal lobes. LP opening pressure was 38 cm H2O, with fluid analysis consistent with hemorrhage, glucose of 53 mg dL−1, and protein of 246 mg dL−1. The patient remained on vancomycin, piperacilin-tazobactam, and voriconazole. Serum and cerebrospinal fluid (CSF) galactomannan antigen by enzyme-linked immunoassay were both negative. On D+37, bronchoscopy showed clear airways without alveolar hemorrhage. Bronchoalveolar lavage cultures and cytology revealed no evidence of mold organisms. Candida glabrata was isolated in culture and was shown to be resistant to all tested azole drugs. Transesophageal echocardiogram was unremarkable for vegetation or shunt.

ATRA was discontinued. The patient's course was further complicated by ventilator dependence, abdominal abscess requiring drainage, continued transfusions for ongoing coagulopathy, oliguria, and probable sepsis while on antibacterial antimicrobial therapy.

On D+45, an esophagogastroduodenoscopy showed a large bleeding duodenal vascular mass.

On D+51, a gastrointestinal (GI) tract perforation was suspected when nasogastric tube feeds began draining through the patient's abdominal abscess drain. A CT of the abdomen and pelvis showed a phlegmon in the left hemiabdomen containing small bowel loops and hematoma in the right hemiabdomen (Fig. 4A) as well as small bowel loops with wall thickening (Fig. 4B). Exploratory laparotomy was remarkable for necrotic bowel involving multiple loops of small bowel with transmural necrosis and perforation. The patient was transitioned to comfort cares and died on D+52.

Figure 4.

A: Coronal reconstructed image of a contrast enhanced abdomen and pelvis CT demonstrates a phlegmon (arrow) in the left hemiabdomen containing small bowel loops and layering blood products. Hematoma (*) in the right hemiabdomen. B: Axial images demonstrate small bowel loops with wall thickening (arrow) within this phlegmon. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Limited postmortem exam without central nervous system (CNS) examination was performed. Hemorrhagic necrosis of the distal duodenum and proximal jejunum without perforation was found, but no convincing invasive fungal organisms consistent with Aspergillus species were identified at autopsy despite very liberal tissue sampling for histopathology. The lungs were remarkable only for mild pulmonary edema and congestion. No convincing leukemic promyelocytes were seen at autopsy; however, our evaluation was limited due to the fact that granulocytes have autolytic changes at autopsy.


APL today is considered to be the most curable subtype of all AMLs, with complete remission (CR) rates and cure rates in >90% and ∼ 80%, respectively [1]. Despite significant improvement in APL outcome, early deaths remain an obstacle during induction phase [2]. APL specific complications such as coagulopathy [1-4] and ATRA syndrome [5, 6] as well as nonspecific complications, including infections due to prolonged neutropenia are responsible for inductions deaths. The infectious diseases society of America (IDSA) recommends antibiotic prophylaxis using a fluoroquinolone in high-risk patients, i.e., those leukemia patients with expected neutropenia >7 days [7]. Consistently, AML patients who have short neutropenia duration (e.g., postconsolidation) do not benefit from antifungal prophylaxis [8]. These guidelines recommend antifungal prophylaxis against Candida infections in patients undergoing intensive remission-induction or salvage-induction for acute leukemia, and antiviral prophylaxis for herpes simplex virus (HSV) in patients undergoing leukemia induction therapy. Other guidelines have recommended use of prophylaxis against invasive Aspergillus (IA) among AML patients [7, 9-11]. Longer neutropenia is an important risk factor for IA because once Aspergillus germinates, host neutrophils are typically the dominant host defense. IA exploits hosts who are neutropenic (quantitative phagocyte defect) or who have other deficiencies in immunity (cell mediated immunity or qualitative deficiencies in phagocyte function) [12].

Although IA can involve almost any organ in an immunocompromised patient, GI tract aspergillosis, either as an isolated (the digestive tract may represent a portal of entry for Aspergillus species in immunocompromised patients) or a disseminated event, is rarely reported. In a 2006 appraisal of the literature, 1,538 cases of aspergillosis were reviewed. Most of these cases predated the use of newer antifungal drugs. Of the IA cases, 5.5% involved the GI tract, with an overall 0.8% having isolated GI tract involvement [13]. This study reported 10 GI tract IA cases in details. Nine of the 10 cases had hematologic malignancy. All underwent laparotomy and small bowel segmental resection due to symptoms similar to typhlitis and peritonitis while receiving antibacterial agents [13]. Kazan et al. reported 21 GI tract aspergillosis; 8 were isolated while 13 had disseminated disease [14]. The most common symptom was abdominal pain (n = 17) followed by diarrhea (n = 10), GI tract hemorrhage (n = 7), intestinal occlusion (n = 6), and perforation (n = 1). Four of the eight patients with isolated GI tract IA had positive serum galactomannan antigen tests. Imaging was performed in 14 of the 21 cases, with 12 of the 14 demonstrating GIT abnormalities. The diagnosis of GI tract IA remains a clinical challenge. Abdominal symptoms and signs are common after chemotherapy and nonspecific for IA. Moreover, there are no characteristic imaging features of GI tract aspergillosis. As in our patient and the patients reported in the literature, diagnosis of GI tract IA requires tissue with an abdominal surgery. After induction chemotherapy, in neutropenic and most likely febrile patients, invasive tests/measures and surgical interventions are avoided unless the risk of mortality is very high otherwise. Therefore, GI tract IA can be missed or possibly misdiagnosed and/or underestimated in AML patients [13, 14]. GI tract can also be involved in disseminated IA although disseminated aspergillosis seems to be decreasing with introduction of mold-active azole and echinocandin drugs to clinical practice. There can be signs/symptoms of other affected organs in these patients, which may assist to make diagnosis of IA. Symptoms of sino-pulmonary system (the most common site of infection) include facial pressure, nasal congestion, facial pain, chest pain, cough, and hemoptysis. CT imaging of the lungs can demonstrate nodular disease with or without cavitation. Haziness that surrounds a nodule, described as a “halo sign,” is a characteristic CT feature of pulmonary infection that results from the immune response to a (likely angioinvasive) organism [15]. Many patients with IA have negative cultures. Even among those patients with radiographic pulmonary abnormalities, bronchoalveolar lavage cultures have ∼ 50% sensitivity [15]. The CNS can be involved in disseminated aspergillosis as well. Clinical symptoms of CNS IA may include seizures or focal neurologic signs. CT or MRI of the head is useful for IA diagnosis [16]. Our patient had CNS symptoms that may represent IA in CNS; however, there was no evidence in sinus, head CT imaging studies, or CSF evaluations. Other diagnostic tools that can be useful in diagnosis of IA are serologic tests. A common biomarker in clinical use for IA is the galactomannan assay. Galactomannan is an Aspergillus cell wall component that is released during Aspergillus growth. The assay can be applied as a surveillance marker, or as a direct diagnostic tool. The cutoff for a positive test differs by (a) the method used to measure the analyte and (b) the reference ranges established in the laboratory performing the test. In one meta-analysis, the galactomannan assay had a sensitivity of 0.71 and a specificity of 0.89 for proven cases of IA [17]. False positives have been caused by concomitant use of certain beta-lactam antibiotics, most notably piperacilin-tazobactam. Sensitivity of the assay is decreased by the concurrent use of anti-fungal agents. The beta-D-glucan assay, detecting 1,3-beta-D-glucan, a cell wall constituent of many fungal organisms, is also used to diagnose invasive fungal infections (IFI). Its overall sensitivity and specificity were 76.8 and 85.3%, respectively [18].

Guidelines for the treatment of aspergillosis have been published by the IDSA [19]. Voriconazole is first line therapy for IA. It has been shown to be more effective than amphotericin B, as initial therapy, with improved survival of 71% vs. 58%, respectively, in a study of 277 patients [20]. There are less data available regarding posaconazole as first line therapy for IA; however, its use is limited by its oral only formulation and the need for a high-fat meal to aid in absorption. Echinocandins are considered salvage therapy [21]. There has been interest in combination therapy involving an echinocandin with either amphotericin B or azole therapy. However, these combinations have demonstrated neutral to synergistic activity in vitro. Data regarding use of combination therapy from a large randomized trial are undergoing evaluation and not available at this time. Further studies may be required to assess the benefit of combination antifungal therapy in IA.

Although we did not have convincing histopathological evidence after autopsy in this case, it is conceivable that IA contributed the patient's mortality. IA has a poor prognosis. Tissue damage in IA results from conidial germination progressing to invasive hyphae, unrestricted growth due to the immune defect, and the following resultant immune response. In the study comparing voriconazole versus amphotericin, which involved 277 patients, 98 died in the first 12 weeks from all causes of mortality, with 56 of the deaths caused by aspergillosis [22]. The poorest prognosis occurs in patients with extrapulmonary aspergillosis. In another case series of 21 patients with GI tract IA, of the 13 patients with disseminated IA, 10 had expired by 12 weeks [14].

This case shows us that although APL is curable, the induction phase of APL treatment is associated with significant mortality as reported by Tallman's group [2]. Most APL mortalities during induction do not result from chemotherapy resistant disease but from complications. This case also highlights that IA of GI tract should be kept in the differential diagnosis of worsening abdominal symptoms and clinical performance during the induction therapy among APL patients with continuing neutropenic fever and no documented microorganism. Diagnosis of IA of the GI tract is most likely underestimated, particularly among patients with no systemic evidence of IA, because its rarity puts it at a lower rank on the differential diagnosis. Moreover, a diagnosis of IA of the GI tract necessitates tissue obtained by invasive surgical procedures that are often avoided in neutropenic patients. The prognosis of IA of the GI tract remains poor, due in part to late diagnosis in the course of infection. Therefore, surgical interventions for diagnosis and treatment should not be delayed in such patients. Antifungal therapy with activity against molds should be employed as soon as possible. Given the fact that mortality in IA is still high in long-term neutropenic patients (e.g., APL induction), perhaps the most critical measure is to prevent IA with appropriate prophylactic agents. This is particularly important in APL, a highly curable leukemia.