Highly Multiplexed Immunohistochemistry Can Predict Steroid Refractory Gastrointestinal (GI) Acute Gvhd at the Time of Endoscopy

Background: For patients with high-risk hematologic disease, hematopoietic cell transplant (HCT) often represents the only option for cure. However, HCT is fraught with complications, leading to morbidity and mortality. Acute GVHD (aGVHD), specifically steroid refractory (SR) GI aGVHD, is a major cause of early mortality. Two issues with SR GI aGVHD predominate: (1) We do not understand the mechanisms differentiating SR vs steroid sensitive (SS) GI aGVHD; (2) We do not have an accurate test to predict SR vs SS GI aGVHD at the time of endoscopic evaluation. Here, we used state-of-the-art multiplexed immunohistochemistry (IHC) to interrogate immune mediators of GI aGVHD, and to develop a robust assay capable of predicting SR vs SS disease from standard GI biopsies.

Methods: We analyzed all HCT patients at Boston Children’s Hospital (BCH) from 1/2014-12/2021 (n=551) who received a first lower GI endoscopy for suspected GI aGVHD (n=109). We tracked their SS vs SR status and their Overall Survival (OS) through 2 yr post-HCT. Classic histology was performed and in addition, each sigmoid colon biopsy (formalin-fixed) was analyzed with a series of novel multiplexed IHC panels (interrogating Pan-CK, DAPI, CD3, CD8, CD20, FoxP3, GITR, CD68, Granzymes B and K, CD103, Ki67, ECP, NCR1, BCMA, CD11c, and MPO, with quantitative image analysis performed using HALO Indica Labs Hyperplex module). Density/proportions of multiple cellular infiltrates and their ratios were assessed. Receiver Operator Characteristic (ROC) and K-nearest neighbors (KNN) classification were used to distinguish SS from SR aGVHD.

Results: Of 109 biopsies, 60 had histologic evidence of lower GI aGVHD inflammation, 12 had other GI inflammation (eg infections, drug effects) and were not further analyzed, and 15 did not have inflammation and therefore were not diagnosed with histologic aGVHD (termed ‘No Inflammation’ (NI). In an additional 22 biopsies, histology was characterized as “Rare Apoptosis” (RA), a diagnosis for which there is no national consensus about severity or the need for steroid treatment. At BCH, the vast majority of patients with RA on endoscopy were treated as not having aGVHD (with only 3/22 receiving steroids, and all 3 being SS). We tracked OS and IHC characteristics for SS aGVHD, SR aGVHD, NI and RA. Fig 1 shows 2 yr OS, which confirms the inferior OS of SR vs SS aGVHD (SR: 58% vs SS: 93% p <0.0001). Importantly, Fig 1 demonstrates that even without steroids, the OS of RA was equivalent to no GI aGVHD (NI: 89% vs RA: 86% p = 0.74).

Multiplexed IHC (performed on 68 biopsies from sequential cases collected from 1/2018-12/2021) was associated with several novel findings (Fig 2): (1) By IHC, RA had fewer immune infiltrates vs aGVHD and NI: CD3 density: aGVHD (both SS and SR combined): 782 ± 133, NI: 386 ± 37, RA: 165 ± 26, p = 9.5x10^-5; CD8 density: aGVHD: 381 ± 83, NI: 123 ± 24, RA: 42 ± 26, p<0.001; CD4 conventional T cells (CD4 Tconv): aGVHD: 359 ± 47, NI: 237 ± 23, RA: 114 ± 19, p<0.001. Ratios of immune infiltrates also distinguished the 3 subtypes: the ratio of CD8:CD4 Tconv: aGVHD: 0.95 ± 0.14; NI: 0.59 ± 0.10 RA: 0.44 ± 0.29 p = 0.013; CD8:CD68: aGVHD: 0.35 ± 0.03, NI: 0.26 ± 0.04 RA: 0.11 ± 0.02, p<0.001. CD8:Epithelial cells (epi): aGVHD: 0.153 ± 0.04; NI: 0.045 ± 0.01; RA: 0.014 ± 0.003, p<0.001. These results underscore the immunologically benign nature of RA, predicting its superior OS.

Importantly, multiplexed IHC could distinguish SR vs SS aGVHD at the time of endoscopy, a finding with major clinical implications. For example, amongst other distinguishing IHC characteristics, the ratio of CD8:CD4Tconv: SR: 1.3 ± 0.22 vs SS: 0.68 ± 0.15 p< 0.01; CD8:epi: SR: 0.220 ± 0.008 vs SS: 0.106 ± 0.042 p = 0.04). A KNN-based classifier identified thresholds with high accuracy (81%), specificity (69%), sensitivity (93%); AUC = 0.852 for SS vs SR aGVHD (Fig 2).

Conclusions: We used state-of-the-art multiplexed IHC to interrogate the immune cells associated with GI aGVHD, with 2 major findings: We find RA associated with superior OS without steroid treatment, and with minimal immune infiltration, confirming its benign nature. We also created a novel multiplexed IHC assay capable of accurately classifying SR vs SS aGVHD from standard-of-care biopsy specimens. If confirmed in a multicenter setting, this would be a major advance, allowing physicians to tailor appropriate therapies to patients at highest risk of developing SR GI aGVHD at the time of initial endoscopic evaluation.