The sensitivity and specificity of Gram stain on formalin fixed tissue sections in the diagnosis of bacterial skin infection - a study of diagnostic test accuracy

This project represents a study of diagnostic test accuracy, aiming to evaluate the test performance (sensitivity, specificity, predictive values, and likelihood ratios) of Gram stain - when performed as part of histological assessment - in the diagnosis of bacterial skin infection. The work-up of patients with a cutaneous disease (e.g. patients with an ulcer, a pustular eruption, or panniculitis) will sometimes involve punch or incisional biopsies wherein part of the tissue is placed in formalin for histological assessment, and part placed in a sterile transport medium (e.g. saline) for microbiological assessment. Microbiological assessment of fresh tissue specimens involves microscopy and culture with or without assessment of antibiotic sensitivities. Histological assessment involves formalin fixation which, in addition to preserving tissue architecture and cell morphology, is bactericidal. After fixation, the tissue is dehydrated, infiltrated with paraffin, and embedded in a rectangular paraffin block. 0.004 mm slices of tissue (so-called tissue sections) are taken from the surface of this paraffin block, stained with haematoxylin and eosin (H&E), and examined at a microscope. Gram stain has been used in in anatomical pathology laboratories in the assessment of formalin-fixed paraffin-embedded (FFPE) tissue sections since the 1880s, but appears to have been accepted as an ancillary test in anatomical pathology laboratories with a relative lack of objective data on test performance to inform its use and interpretation. Gram stain can either be performed as on fresh tissue specimens as part of microbiological assessment or on formalin fixed paraffin embedded (FFPE) tissue as part of histological assessment. Under both circumstances, Gram stain involves the use of a primary stain (crystal violet) and an iodine solution which bind to bacterial nucleic acids, followed by a clearing agent (alcohol or acetone) and a counterstain (e.g. safranin). Gram positive bacteria appear purple due to retention of the crystal violet-iodide complex, a result of their cell wall structure particularly in terms of the amount and configuration of peptidoglycan, whereas Gram negative bacteria are decolourised by the clearing agent and appear red due to the counterstain. Although Gram stain plays an important role in microbiological assessment in terms of visualisation and classification of bacteria, its use on FFPE in histological assessment has multiple limitations which include: the time intensive nature of searching for small numbers of bacteria on stained sections, the lack of visual contrast between bacteria (particularly Gram negative bacteria) and background tissue, and the ambiguity in distinguishing between bacterial colonisation and genuine infection. The potential advantages of using Gram stain on tissue sections include: the potential value of visualising bacteria in a particular tissue plane (e.g. deep dermis, blood vessel wall) to distinguish genuine infection from colonisation (where bacteria tend to be limited to the surface of the specimen), the identification of bacteria which are slow or difficult to culture (e.g. B. henselae), and in attempting to determine the pathogenic organism when culture reveals a combination of Gram positive and negative bacteria. Notwithstanding, there is a relative lack of objective data pertaining to the diagnostic test accuracy of using Gram stain on FFPE tissue sections to diagnose infection. Guarner et al (2007) evaluated the test performance of Gram stain and immunohistochemistry in diagnosing S. pneumoniae infection, using tissue culture as the reference standard (n = 46). In this context, Gram stain had sensitivity of 0.61 (61%) and specificity of 0.50 (50%) in diagnosing S. pneumoniae infection. These results suggest that the test performance or diagnostic accuracy of Gram stain on tissue sections is poor, at least in the diagnosis of S. pneumoniae. This project aims to add to existing literature by using a cohort with a large sample size to evaluate the test performance of Gram stain on FFPE tissue sections (index test) in diagnosing bacterial skin infection. These aims will be addressed by reviewing a reviewing case notes and pathology reports as part of a retrospective and consecutive series of cases (with a time interval spanning September 2017 to the date on which data collection begins) wherein skin tissue was concurrently submitted for microbiological and histological assessment at SA Pathology. Cases will be excluded if the Gram stain result was not described in the anatomical pathology report, or if the patient was diagnosed with bacterial infection due to a pathogen which cannot reasonably be identified using Gram stain (e.g. mycobacteria). For each case, medical records including case notes available through SA Health, microbiology reports, histopathology (anatomical pathology) reports, and other test results (e.g. white cell count, c-reactive protein) will be reviewed. This project will not involve active participant participation, as the data to be reviewed (case notes and test results) will be that which was collected as part of the participant’s routine medical care. A waiver of consent has been granted in line with the NHRMC National Statement on Ethical Conduct in Human Research 2007 (Updated 2018). For each case, the duration of observation will be encompassed by the review of medical records spanning the patient’s first presentation to the hospital system (with the cutaneous condition being investigated) until the point at which a working diagnosis was established and treatment initiated. For each case, relevant medical records are expected to span a one to two-week period. The performance of the index test will then be evaluated by comparing the Gram stain results as described in anatomical pathology reports with tissue microscopy and culture results and the final diagnosis assigned to each patient’s cutaneous disease. The reference standard will be defined as the final diagnosis assigned to each patient’s cutaneous disease, determined through a process of clinicopathological correlation (i.e. the reconciliation of clinical and pathological information).