Discrepancies Between Clinical and Anatomic Causes of Death in Acute Invasive Fungal Sinusitis: An Institutional Experience at a Tertiary Care Center

After institutional review board review with exemption, the medical and surgical records of 9 patients with a pathologic diagnosis of AIFS at autopsy were retrospectively reviewed, from January 1, 2000, through March 19, 2020. Inclusion in the study required that (1) the autopsies were not restricted to a single organ or body cavity, (2) the patients had a hospital stay of more than 24 hours, and (3) the patients had evidence of invasive fungal organisms in histologic sections taken at autopsy. All autopsy reports with the terms invasive fungal sinusitis, fungal cerebritis, Mucor, Aspergillus, and Rhizopus were reviewed by the authors. Clinical information was obtained from a review of the electronic medical records and included discharge summaries, operative reports, radiologic reports, and laboratory and pathology reports. Data collection included patient age at the time of death, sex, ethnicity, clinical diagnoses known at the time of death, and pathologic findings from autopsy.

A total of 9 cases were reviewed for discrepancies between the clinical and anatomic causes of death in patients with AIFS, determined by the presence of invasive fungal elements within the brain parenchyma on histopathologic section. Discrepancy between the clinical and autopsy diagnoses was assessed based on the modified Goldman classification. Discrepancies can be categorized into 6 classes, with classes I and II considered major discrepancies and classes III and IV minor discrepancies. Class V is used when there is complete agreement between the clinical and pathologic diagnoses, and class VI is used when a case is not able to be classified (Table 1).

From January 1, 2000, through March 19, 2020, a total of 7071 autopsies were performed at our institution. Acute invasive fungal sinusitis was diagnosed in 9 autopsy cases (0.13% of all autopsies). The study cohort included 4 women and 5 men, and the mean age at autopsy was 42 years (range, 8–70 years). All 9 patients had an underlying disease process that caused them to be immunocompromised. Underlying diagnoses included hematologic malignancy (4 of 9; 44%), chronic steroid use (3 of 9; 33%), poorly controlled diabetes (1 of 9; 11%), and active hepatitis C infection (1 of 9; 11%). A total of 3 of the 4 patients with hematologic malignant neoplasms had undergone bone marrow transplantation (Table 2, cases 3, 5, and 8). Prolonged steroid use was also observed in this cohort of autopsy cases, and 3 patients received a course of prednisone-equivalent steroids for at least 2 weeks' duration (Table 2, cases 2, 7, and 9).

Of the nine cases that met inclusion criteria, 33% (n = 3) demonstrated clinical onset of symptoms prior to hospitalization. The most common presenting symptoms were fever (7 of 9; 78%), face pain (3 of 9; 33%), and ophthalmologic complaints (2 of 9; 22%), with headache occurring less frequently (1 of 9; 11%). Many patients (7 of 9; 78%) had 2 or more of these symptoms, and the remaining 2 patients presented with fever alone. The mean time from admission to time of expiration was 29 days (range, 3–64 days).

For all 9 autopsy cases, blood culture, including evaluation of fungal organisms, was performed during hospitalization. Antemortem bronchoalveolar lavage was performed in 2 cases (Table 2, cases 2 and 7). For case 2, the clinical cause of death was pneumonia and septic shock, and the clinical question to be addressed by the autopsy was to ascertain the cause of sepsis and hypoxia. For case 7, clinical cause of death was a bilateral pneumothorax and lower gastrointestinal tract bleed. The clinical question to be addressed by the autopsy focused on the lungs in both cases, with the first focused on the cause of hypoxia and the second focused on the pulmonary pathology.

Other laboratory evaluations in this patient cohort included antemortem tissue biopsy and frozen section analysis. For case 4, computed tomography of the brain demonstrated ill-defined, hypoattenuating left frontal lobe lesions that were biopsied and demonstrated rare fungal organisms, with morphologic features consistent with Zygomycetes species. The clinical cause of death was encephalitis secondary to mucormycosis. Sinus debridement and intraoperative frozen section were performed in 1 case (Table 2, case 6). A head computed tomography scan demonstrated opacification of the right maxillary sinus (Figure 1, A), and subsequent debridement and intraoperative frozen section demonstrated fungal hyphae in the background of necrotic and inflammatory debris (Figure 1, B). Following the sinonasal debridement, the patient developed hemorrhagic conversion of previous ischemic changes, had worsening cerebral edema, and died 2 days later. At autopsy, intravascular thrombi within the cavernous and sphenoid sinuses were grossly identified (Figure 2, A), and angioinvasive fungal hyphae consistent with Mucor were microscopically demonstrated throughout the brain parenchyma (Figure 2, B).

In this study cohort, angioinvasive fungal elements were identified within the brain parenchyma in all cases, with multiorgan involvement in 67% of cases (n = 6). Using the modified Goldman criteria, there was complete agreement between the clinical and autopsy diagnoses in 3 cases, where fungal sinusitis was listed as a cause of death in 2 cases and encephalitis secondary to mucormycosis in 1 case (Table 2, cases 1, 4, and 6). In this study, major discrepancies between clinical and autopsy diagnoses related to AIFS were identified in 6 cases (Table 2, cases 2, 3, 5, 7, 8, and 9). A total of 67% (4 of 6) of the major discrepancy cases were Goldman class I, where angioinvasive fungal sinusitis was not diagnosed clinically and had an adverse impact on survival.

The present study is a retrospective review of all complete pediatric and adult clinical and autopsy reports for the past 20 years at a regional referral center in which there was a histologic diagnosis of AIFS. It excluded cases from outside institutions, patients with short hospital stays, and those in which the autopsy was restricted to a single organ in order to draw a meaningful correlation between clinical and pathologic causes of death. Although the exclusion criteria made the size of the study smaller, the entire set of included cases was comprehensively reviewed, and it included pediatric data, which previously has not been extensively studied in the context of AIFS and autopsy.

There are several limitations to this study. It is a retrospective review, necessitating review and interpretation of clinical notes not written in a standardized fashion. Therefore, there is a certain degree of subjectivity in classification of missed diagnoses; however, we addressed this by having multiple reviewers for each case. Additionally, our institution uses a standardized form completed at the time of death, with a clinical cause of death provided by the clinical team, which also mitigates the subjectivity in interpreting the clinical cause of death. Another important consideration in this present study is that there is a selection bias in that not all deaths in our hospital have a postmortem examination performed. If the study included performing autopsies on all hospital deaths, it is likely that incidence of discrepant diagnoses for AIFS may have increased in the setting of occult, nonprototypical clinical presentation of disease.

Another significant limitation of this study is that sinus and nasal sections were not extensively sampled at the time of autopsy in this study cohort. Angioinvasive fungal organisms were identified in the brain parenchyma of all cases, with evidence of multiorgan involvement in a subset of cases, suggesting that fungal sinusitis was likely present, but only histologically confirmed in 1 case (Table 2, case 6). The pathogenesis of invasive fungal sinusitis has not been fully elucidated. Previous studies have suggested that most cases of invasive fungal sinusitis begin within the nasal cavity, with a propensity for the middle turbinate, and then invade deeper structures, such as the sinuses and cranial vault as the disease progresses.¹⁵  For case 6, where there was antemortem frozen section confirming AIFS, and the clinical cause of death was documented as fungal sinusitis in the setting of poorly controlled diabetes, our postmortem examination included a nonstandard autopsy section that blocked out the entire pituitary region with the bone to include the cavernous and sphenoid sinuses and a portion of the internal carotid artery. It should be noted that removing a portion of the internal carotid artery creates challenges in the embalming process. It is our recommendation that if such a section is taken at autopsy, this should be clearly communicated to the funeral home that will be receiving the body to ensure optimal interprofessional teamwork in the care of human remains.

Acute invasive fungal sinusitis is an urgent and life-threatening infection commonly associated with an immunocompromised status, either from treatment for an underlying malignancy, from treatment with steroids, or from diseases such as diabetes mellitus.^16–19  In the present study, all patients had a clinical history of immunosuppression (Table 2). Given the broad spectrum of clinical presentations that has been previously reported and what we observed in our study cohort, antemortem diagnosis of AIFS can be challenging, especially in cases where patients do not present with prototypical signs and symptoms of disease. Thus, as demonstrated in previous studies, a high degree of clinical suspicion should be exercised in patients with underlying immunosuppression or hematologic malignancy. Prior studies have proposed algorithms for timely identification and treatment of AIFS patients.¹⁵  This algorithm includes proper identification of at-risk populations, such as those with an underlying immunocompromised state with associated risk factors such as heavy steroid use, in the setting of fevers of unknown origin with localizing symptoms, such as facial pain. This clinical scenario should prompt immediate evaluation either with rigid nasal endoscopy or axial/coronal computed tomography scan for evidence of sinusitis.¹⁵  We note that in 2 of the major discrepant cases (Table 2, cases 2 and 7) neurologic imaging was not performed, which deviates from the proposed clinical algorithm and may have contributed to the adverse outcomes in these 2 cases. Although there have been extensive advances in modern antemortem clinical investigation, autopsies have continued to reveal major antemortem diagnostic errors in as many as 30% of cases.^11–14  In our present study, there was a discordance between the clinical and anatomic causes of death in 67% of cases (6 of 9 cases), with most cases identified as class I discrepancies. In the major discrepant cases, classes I and II, an underlying infectious process was clinically suspected in 4 of the 6 cases; however, fungal etiology was not correctly identified. All cases in this study cohort demonstrated fungal organisms within the brain parenchyma, with a subset of cases having multiorgan involvement. This observation supports the utility of extensive tissue sampling to include sampling of sinus and nasal sections in clinically equivocal cases and may lend new insights into disease etiology. Autopsies should continue to be the gold standard for quality control in clinical management, including radiologic evaluation,^20–25  in spite of declining requests.²⁴  The most crucial factors influencing attitudes toward the autopsy have been shown to be the clinician's level of experience with autopsy in training and practice.²⁴  The importance of autopsies should be emphasized in medical education and postgraduate training as a measure of decreasing clinical and diagnostic discrepancies, ultimately improving and optimizing patient care.^20–25  Our study highlights the importance of the pathologist's role at autopsy in the definitive diagnosis of AIFS, especially in clinically equivocal cases.

We would like to thank Mr Orlando Crisp, Mr Jessie Hardges, and Mr Byron Henderson for their technical expertise with the autopsy prosections.