Patient-derived epithelial cell organoids mimic the phenotypic complexity of endometriosis subtypes.

Establishment of a Patient-Derived Organoid Culture Platform for Mantle Cell Lymphoma Drug Sensitivity and CAR T Cell Anti-Tumor Activity Assay

Matrix metalloproteinase 7 expression of endometrial epithelial cells: a potential marker for patients with deep infiltrating endometriosis?

Background: Patient-derived xenografts (PDX) and organoids (PDO) have become important translational model systems for cancer research, especially for drug responses prediction. Since PDX models have limitation as low success rate and takes long time, PDO is emerging as a new technique. Here, we compared the drug responses in PDX and PDO models using tissues obtained from treatment-refractory breast cancer patients. Methods: Tumor tissues from breast cancer patients were implanted into the mammary fat pads of immunodeficient mice. Tumor size of mice were measured 3 times a week using digital caliper. When xenograft tumors reached 200mm3 in size, drug treatment was started. Drugs selection was based on gene expression patterns, the presence of available drugs, and clinical treatment history. The organoids were established from PDX tumor pieces. Organoids were seeded and cultured in 96-well plates (2000 cells per well) for drugs testing. We treated with a single or combination drugs in PDX and PDO models. For interpretation of drug sensitivity results of PDO or PDX, we referred to IC50 database of 2D cell lines or results from references. Results: We compared the drugs response efficacy in five cases with paired PDX and PDO models; 1 hormone receptor (HR) positive+; HER2 negative-, 1 HR+;HER2+ and 3 TNBC (triple-negative breast cancer) subtypes. In HR+;HER2- case, drug test results between PDX and PDO partially matched in single treated group. In HR+;HER2+, the results of combination treated groups were partially matched in PDX and PDO model. The results for two TNBC samples matched in single or combination treated groups. Especially, tumor size or cell viability of one TNBC case showed significant differences between control and sorafenib/everolimus combination treated groups. The other case of TNBC type had partially matched in PDX and PDO model. We will analyze the consistency for the genomic profiles of tumors in patients, PDX, and PDO models. Conclusion: We have compared the various drugs responses through the successful establishment of PDO and PDX from the different breast cancer subtypes. Although the results are not perfectly matched, it showed that these models have potential to assist the chemotherapy strategies for each patient and predict outcome of treated patient's prognosis. In the future, we will be focused on explaining why the results of drug response between PDX and PDO were inconsistent. (This study was supported by National Cancer Center, Korea, 1710450, and 1810101) Citation Format: Jin-Sun Ryu, You-sun Noh, Bo-Ra Kim, Yun-Hee Kim, A-Ra Jeon, Sung Hoon Sim, In Hae Park, Eun Gyeong Lee, Eun Sook Lee, Keun Seok Lee, Sun-Young Kong. Comparison of drug responses using patient-derived xenograft (PDX) and patient-derived organoid (PDO) models from treatment-refractory breast cancer [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1667.

Effects of hormonal treatment on the expression of collagen type I, matrix metalloproteinase-1, and tissue inhibitor of metalloproteinases-1 in endometriosis.

Serous endometrial and ovarian cancers represent significant morbidity and mortality for gynecologic cancers due to the high rate of resistance to chemotherapy and recurrence after treatment. Among the many reasons for poor outcomes is a lack of preclinical models that reflect the complex heterogeneity across patients. Our objective was to create patient-derived organoid (PDO) models of serous endometrial and ovarian cancer and examine how genomic profiles evolve in response to standard therapy. This study was performed in four PDO models of serous gynecologic cancer, including serous endometrial, high grade and low grade serous ovarian, and high grade serous fallopian tube. We first compared genomic alterations in the primary tumors and PDOs using a 484-gene NGS panel (NovoPM 2.0, Novogene). A large overlap in single nucleotide variants (SNVs), copy number variations (CNVs) and indels was observed between primary tumor tissue and the corresponding PDO model. For example, there was an average of 175 shared SNVs between each primary tumor and PDO model and <20 unique variants in the PDO that were not present in primary tumor specimen. For the patient with serous fallopian tube cancer, we were further able to generate PDO models from tumor tissue acquired from three different sites: ovary, omentum, and ascites fluid. We found that 217 SNVs were shared among the PDOs from the three sites, with only 2-11 variants unique to each location. Interestingly, eight unique CNVs were detected in the ovary and ascites PDOs but not in the metastatic (omentum) PDO. We next exposed each PDO to a short 3-day pulse of carboplatin+paclitaxel to create chemoexposed models. The rationale for this duration of exposure is that clinical response in patients has been shown to correlate with drug response at 72 hrs in PDO models of ovarian cancer. As expected, all chemoexposed PDOs were more resistant to chemotherapy as compared to treatment-naïve counterparts. Genomic analysis of the treatment-naïve vs. chemoexposed PDOs revealed acquisition of new variants, such as a p53 mutation in the serous endometrial model after the pulse of chemotherapy. Finally, we compared drug sensitivity of the treatment-naïve vs. chemoexposed PDOs to agents used in the adjuvant and recurrent settings. The chemoexposed models yielded different drug profiles, with some showing increased sensitivity and others increased resistance. Taken together, these data substantiate that PDO models retain tumor heterogeneity, exhibited by the different genomic profiles and varying chemosensitivity. In addition, PDO models can be used to model the genomic and drug response profiles that arise in response to chemotherapy. Citation Format: Andreea Newtson, Emily Symons, Paige Malmrose, Eric Devor, Samantha Parks, Craig Rush, Jessica Andrew-Udoh, Haley Losh, Jay Gertz, Kristina Thiel, Kimberly Leslie. Use of patient-derived organoids to model tumor evolution in response to chemotherapy [abstract]. In: Proceedings of the AACR Special Conference on Endometrial Cancer: Transforming Care through Science; 2023 Nov 16-18; Boston, Massachusetts. Philadelphia (PA): AACR; Clin Cancer Res 2024;30(5_Suppl):Abstract nr B032.

Three types of endometriosis are described: superficial peritoneal endometriosis (SPE), ovarian endometrioma (OMA), and deep infiltrating endometriosis (DIE). The expression of somatostatin receptors (SSTR1, 2, and 5) in human endometrial tissue and its ectopic form has been previously studied and may be different in each type of endometriosis. The aim of this study was to assess the immunohistochemical expression of SSTR1, 2, and 5 in tissue samples of SPE, OMA, and DIE. We performed a retrospective analysis in the pathology department database. Patients aged <50 yr and diagnosed with endometriosis have been identified and sorted into 3 groups according to their endometriosis type: SPE, OMA, and DIE. For each selected patient, formalin-fixed paraffin-embedded blocks were retrieved in order to make new sections to be incubated with polyclonal rabbit antibodies anti-SSTR1, 2, and 5. Receptor status was considered as positive on the sections when >50% of the cells showed immunostaining. Seventy-six patients were included in the analysis. SSTR1 and 5 were expressed in 95.4% and 77.2% of SPE, respectively, in 95.8% and 83.3% of OMA, respectively, and in 100% and 80% of DIE, respectively. There was no significant difference between SPE, OMA, and DIE with regard to the SSTR1 (P=0.5) and SSTR5 (P=0.9) expression. We observed a significant difference between SPE (9.0%), OMA (16.6%), and DIE (63.3%) with regard to SSTR2 expression (P<0.05). The present study identifies 2 different immunohistochemical patterns of endometriosis lesions with regard to their SSTR expression: SSTR1+/SSTR2-/SSTR5+ for SPE and OMA, and SSTR1+/SSTR2+/SSTR5+ for DIE.

Introduction: Predicting response to therapy for each patient’s tumor is critical to improving long-term outcomes for muscle-invasive bladder cancer (MIBC). Our objective is to establish ex vivo bladder cancer patient-derived organoid (PDO) models that are representative of patients’ tumors and determine potential efficacy of standard-of-care and curated experimental therapies. Methods: Tumor material was prospectively collected from consented patients with bladder cancer to generate short-term PDO models, which were screened against a panel of clinically relevant drugs in ex vivo 3D culture. Drugs were tested at the maximum plasma concentration (Cmax) in human trials, so as to provide physiologic relevance. When material was sufficient, cells were additionally tested using full dose response. The number of drugs screened was dependent upon amount of tissue available, with up to 34 drugs screened at Cmax and 9 drugs screened in dose response format. Multiomic profiling was utilized to validate the PDO models, establish the molecular characteristics of each tumor, assess gene expression (GEX) patterns between paired primary tissue and PDOs, and identify potential biomarkers of drug response using Pearson correlation coefficients and Kruskal-Wallis tests with Dunn’s post-hoc pairwise comparison testing. Results: 106 tumors were collected from 97 patients, with 65 samples yielding sufficient material for complete multiomic molecular characterization and PDO screening with 6 to 32 drugs/combinations. RNA sequencing and GSEA enriched pathways across organoids were compared to other published datasets, noting high levels of correlation between our organoid models and in particular the Cancer Cell Line Encyclopedia urethral lines. Moreover, analysis of subtype and mutation/copy number data further indicate that our organoid dataset is representative of the full spectrum of disease. RNA sequencing and subtyping of cultured organoids compared to patient tissue indicate that organoid models are representative of patient tissue and do not undergo subtype shifts in our short-term 3D cultures. Our drug screening results highlight the large spectrum of response to chemotherapies that in fact, clinically benefit only a small proportion of patients in the neoadjuvant setting. Utilizing an integrative approach, novel correlations between ex vivo drug responses and genomic alterations, GEX, and protein expression were identified, including a multiomic signature of gemcitabine response. Conclusions: Rapid organoid development, characterization, and drug screening allows for the prediction of therapeutic response in ~10 days following sample collection. Use of this technique on tissue provided during disease work-up may further guide selection of effective therapeutic agents in patients with bladder cancer, beyond the typical unranked lists derived from genomic analyses. This work would help maximize the overall benefits of standard of care therapies and could be used to identify alternative therapeutics for patients who progress on standard therapies. Citation Format: Nathan M. Merrill, Samuel D. Kaffenberger, Liwei Bao, Nathalie Vandecan, Laura E. Goo, Athena Apfel, Xu Cheng, Zhaoping Qin, Chia-Jen Liu, Armand Bankhead, Yin Wang, Varun Kathawate, Lila Tudrick, Habib Serhan, Zackariah Farah, Chad Ellimoottil, Khaled S. Hafez, Lindsey A. Herrel, Jeffrey S. Montgomery, Todd M. Morgan, Simpa S. Salami, Alon Z. Weizer, Peter J. Ulintz, Mark L. Day, Matthew B. Soellner, Phillip L. Palmbos, Sofia D. Merajver, Aaron M. Udager. Integrative drug screening and multi-omic characterization of patient-derived bladder cancer organoids reveals novel molecular correlates of chemotherapy response [abstract]. In: Proceedings of the AACR Special Conference on Bladder Cancer: Transforming the Field; 2024 May 17-20; Charlotte, NC. Philadelphia (PA): AACR; Clin Cancer Res 2024;30(10_Suppl):Abstract nr A009.

One of the major challenges in preclinical cancer therapeutic development is establishing physiologically relevant in vitro assays that correlate with the in vivo responses of patient tumors to anticancer agents. By incorporating tumor cell heterogeneity and three-dimensional morphological features, patient-derived organoids provide an improved in vivo relevancy compared to established tumor cell lines grown as monolayers. However, organoids grow while embedded in an extracellular matrix material with complex media formulations, which poses challenges for culture scale-up and automated drug screening methods. Organoid models derived from a variety of human solid tumor types are distributed by the National Cancer Institute’s Patient-Derived Models Repository program (https://pdmr.cancer.gov). A panel of human patient-derived colon adenocarcinoma organoid models was assembled to evaluate an automated high-throughput screening (HTS) platform. The organoid panel members were characterized for their reproducible growth and expansion capacity in culture, recovery from cryopreservation, and amenability to operations associated with HTS. Short tandem repeat profiling was performed regularly throughout the process to authenticate each sample. Among the organoid models, variations were observed in morphology (assessed by brightfield imaging) and growth rate (measured by population doublings). Most models expanded well in culture for greater than sixty days and all models demonstrated a sufficient recovery from cryopreservation. The aims in adapting organoid cultures to a HTS platform included minimizing the operational complexity, maximizing the process throughput, and maintaining high organoid viability. Assay conditions for all panel members were selected in conjunction with automated methods, instrumentation, and endpoint measurements. Details such as the optimal sample preparation steps, media formulation, and inoculation density varied among the organoid models. However, other aspects such as liquid handling procedures for organoid inoculation and drug delivery, microwell plate type, assay duration, and endpoint measurements were selected for their suitability to all organoid models tested. Using a custom-designed automated screening system, the refined methods were validated by screening the panel of patient-derived colon adenocarcinoma organoids against a library of oncology drugs approved by the United States Food and Drug Administration (https://dtp.cancer.gov/organization/dscb/obtaining/default.htm). Assay performance metrics and pharmacological data demonstrate the robust performance of this organoid screening platform. Future efforts will establish additional patient-derived organoid panels for expanded HTS using this platform. This project was funded in part with federal funds from the NCI, NIH, under contract no. HHSN261201500003I. Citation Format: Siddhartha Paul, Curtis Hose, Eric Jones, Erik Harris, John Connelly, Petreena Campbell, Mariaestela Ortiz, Thomas S. Dexheimer, Thomas Silvers, Penny Sellers Brady, Julie Grams, Tiffany Nikirk Rohrer, Karen Martin, Patricia Ramsey, Lori Bowles, Annamaria Rapisarda, Ralph E. Parchment, Beverly A. Teicher, James H. Doroshow, Nathan P. Coussens. Development of an automated platform for screening patient-derived organoid models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3079.

Pancreatic ductal adenocarcinoma (PDAC) is amongst the most lethal malignancies. Recently, next-generation organoid culture methods enabling the 3-dimensional (3D) modeling of this disease have been described. Patient-derived organoid (PDO) models can be isolated from both surgical specimens as well as small biopsies and form rapidly in culture. Importantly, organoid models preserve the pathogenic genetic alterations detected in the patient's tumor and are predictive of the patient's treatment response, thus enabling translational studies. Here, we provide comprehensive protocols for adapting tissue culture workflow to study 3D, matrix embedded, organoid models. We detail methods and considerations for isolating and propagating primary PDAC organoids. Furthermore, we describe how bespoke organoid media is prepared and quality controlled in the laboratory. Finally, we describe assays for downstream characterization of the organoid models such as isolation of nucleic acids (DNA and RNA), and drug testing. Importantly we provide critical considerations for implementing organoid methodology in a research laboratory.

Organoid models have recently been utilized to study 3D human-derived tissue systems to uncover tissuearchitecture and adultstem cellbiology. Patient-derived organoids unambiguously provide the most suitable in vitro system to study disease biology with the actual genetic background. With the advent of much improved and innovative approaches, patient-derived organoids can potentially be used in regenerative medicine. Various human tissues were explored to develop organoids due to their multifold advantage over the conventional in vitro cell line culture approach and in vivo models. Gastrointestinal (GI) tissues have been widely studied to establish organoids and organ-on-chip for screening drugs, nutraceuticals, and other small molecules having therapeutic potential. The function of channel proteins, transporters, and transmembrane proteins was also explained. The successful application of genome editing in organoids using the CRISPR-Cas approach has been reported recently. GI diseases such as Celiac disease (CeD), Inflammatory bowel disease (IBD), and common GI cancers have been investigated using several patient-derived organoid models. Recent advancements on organoid bio-banking and 3D bio-printing contributed significantly in personalized disease management and therapeutics. This article reviews the available literature on investigations and translational applications of patient-derived GI organoid models, notably on elucidating gut-microbial interaction and epigenetic modifications.

BackgroundTo investigate the characteristics of deep infiltrating endometriosis (DIE) lesion distribution when associated with ovarian endometrioma (OEM).MethodsThe present study analyzed retrospective data obtained by the First Affiliated Hospital of Sun Yat-sen University, between June 2008 to June 2016. A total of 304 patients underwent laparoscopic surgery for complete removal of endometriosis by one experienced surgeon, and histological confirmation of OEM associated with DIE was conducted for each patient. Clinical data were recorded for each patient from medical, operative and pathological reports. Patients were then divided into two groups according to unilateral or bilateral OEM. Patients with unilateral OEM were subsequently divided into two subgroups according to OEM location (left- or right-hand side) and the diameter of the OEM (≤50 and > 50 mm). The distribution characteristics of DIE lesions were then compared between the groups.ResultsDIE lesions were widely distributed, 30 anatomical sites were involved. Patients with associated unilateral OEM (n = 184 patients) had a significantly increased number of DIE lesions when compared with patients with bilateral OEM (n = 120 patients; 2.76 ± 1.52 vs. 2.33 ± 1.34; P = 0.006). Compared with bilateral OEM with DIE, there was a higher rate of intestinal (39.1% vs. 18.3%; P < 0.01) and vaginal (17.4% vs. 6.7%; P < 0.01) infiltration by DIE lesions in unilateral OEM with DIE. The mean number of DIE lesions was not significantly correlated with the location or size of the OEM (2.83 ± 1.56 vs. 2.74 ± 1.53; P = 0.678; and 2.65 ± 1.42 vs. 2.80 ± 1.43; P = 0.518, respectively). There was no significant difference between the groups with OEM ≤50 mm and > 50 mm.ConclusionLesion distribution characteristics in women diagnosed with histologically proven OEM associated with DIE were frequently multifocal and severe.

Background: The National Cancer Institute (NCI) has developed a Patient-Derived Models Repository (PDMR; https://pdmr.cancer.gov) of preclinical models including patient-derived xenografts (PDX), organoids (PDOrg) and patient-derived cell cultures (PDC). Extensive clinical annotation and genomic datasets are available for these preclinical models. However, it is unclear if the molecular profiles of the corresponding patient tumors are stably propagated in these models. We have previously demonstrated that PDX models from the NCI PDMR faithfully represent the patient tumors both in terms of genomic stability and tumor heterogeneity. Here, we conduct an in-depth investigation of genomic representation of patient tumors in the PDOrgs and PDCs. Methods: PDOrgs (n=64) and PDCs (n=94) were established from tumor fragments (i.e., initiator specimens) obtained either from patient specimens or from PDX specimens of early passage. For some models (n=19), both PDOrgs and PDCs were generated from the same tumor tissue; in fewer cases (n=4), PDCs were established from organoids derived from patient specimens. Whole Exome Sequencing and RNA-Seq were performed on all PDCs and PDOrgs, and data were compared with patient specimens or early passage PDXs. Results: A majority of the PDOrgs and PDCs have stably inherited the genome of the corresponding patient specimens based on the following observations: (1) &amp;gt;87% of PDOrgs and PDCs maintained similar copy number alteration profiles compared with the initiator specimens of the preclinical model; (2) the variant allele frequency (VAF) of clinically relevant mutations remained consistent between the PDOrgs, PDCs, and the initiator specimens, with none of the PDCs or PDOrgs deviating by &amp;gt;15% VAF; and (3) clinically relevant biomarkers (e.g., MSI, LOH, mutational signatures etc.) are concordant amongst the PDOrgs, PDCs, and the initiator specimens. We observed that the majority of SNVs and indels present in the initiator specimens were also found in the PDOrgs and PDCs, suggesting almost all the tumor heterogeneity was preserved in these preclinical models. Conclusions: This large and histologically diverse set of PDOrgs and PDCs from the NCI PDMR exhibited genomic stability and faithfully represented the tumor heterogeneity observed in corresponding patient specimens. These preclinical models thus represent a valuable resource for researchers interested in pre-clinical drug or other studies. Citation Format: Biswajit Das, Yvonne A. Evrard, Li Chen, Rajesh Patidar, Tomas Vilimas, Justine N. McCutcheon, Amanda L. Peach, Nikitha V. Nair, Thomas D. Forbes, Brandie A. Fullmer, Anna J. Lee Fong, Luis E. Romero, Alyssa K. Chapman, Kelsey A. Conley, Robin D. Harrington, Shahanawaz S. Jiwani, Peng Wang, Michelle M. Gottholm-Ahalt, Erin N. Cantu, Gloryvee Rivera, Lindsay M. Dutko, Kelly M. Benauer, Vishnuprabha R. Kannan, Carrie A. Bonomi, Kelly M. Dougherty, Joseph P. Geraghty, Marion V. Gibson, Savanna S. Styers, Abigail J. Walke, Jenna E. Moyer, Anna Wade, Mariah L. Baldwin, Kaitlyn A. Arthur, Kevin J. Plater, Luke Stockwin, Matthew R. Murphy, Michael E. Mullendore, Dianne L. Newton, Melinda G. Hollingshead, Chris A. Karlovich, Paul M. Williams, James H. Doroshow. Patient-derived organoid and cell culture models from the NCI Patient-Derived Models Repository (NCI PDMR) preserve genomic stability and heterogeneity of patient tumor specimens [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 3916.

Endometriosis: Novel Models, Diagnosis, and Treatment

Background Intratumoral heterogeneity plays an important role in glioblastoma (GB) resistance to standard therapy consisting of irradiation and chemotherapy with temozolomide (TMZ). However, classical in vitro GB models fail to represent the complex cellular composition of tumors in vivo, which hinders relevant examination of GB therapeutic response. To overcome these limitations, we studied the effects of irradiation and TMZ in a novel patient-derived organoid model. Material and Methods We established a patient-derived GB organoid model by a protocol recently published by Jacob et al. Original tumor tissue and tissue-derived organoids were compared by immunofluorescence staining of selected cell type markers and qPCR analysis of expression levels of a panel of selected target genes, including 15 genes defining GB subtypes. To analyze GB therapeutic response, organoids from 11 patients were exposed to a single dose of irradiation (10 Gy), one-week treatment with TMZ (50 µM) or their combination. The effects of therapy were assessed by viability and invasion assays. Expression levels of a number of genes related to GB subtypes, epithelial-mesenchymal transition, stemness, DNA damage responses, cell cycle, cytokines, and cell markers of the tumor microenvironment (TME) were compared between treated organoids and untreated controls. In addition, the heterogeneity of the TME and its responses to treatment were investigated by spatially resolved transcriptomics with in situ sequencing (ISS) methodology. Results Organoids recapitulate inter-patient variability and reflect the cellular composition and gene expression levels of the tumor tissue from which they were derived. GB stem cells and differentiated cancer cells are present in organoids along with various cells of the TME, e.g., macrophages and microglia, lymphocytes, and endothelial cells. Irradiation and TMZ showed no significant effects on organoid viability and invasion. However, some target genes were differentially expressed in the treated organoids, such as E3 ubiquitin-protein ligase MDM2 and cyclin-dependent kinase inhibitor 1A (CDKN1A). To our knowledge, we are the first to apply spatially resolved transcriptomics (ISS) to formalin-fixed, paraffin-embedded sections of (un)treated GB organoids. Our results elucidate the role of the TME in GB therapeutic response and shed light on potential mechanism underlying GB therapy resistance. Conclusion Patient-derived GB organoids recapitulate the key characteristics and complex composition of patient’s tumor tissue, providing a valuable platform for studies of GB therapeutic response and resistance.

Recurrence of endometriosis after surgery constitutes a serious challenge. Whether there is an evolution of lesion subtypes with each recurrence and whether certain lesions subtypes tend to recur faster than others is not adequately addressed. Medical records of all patients who underwent surgery for endometriosis between 1997 and 2018 in the Department of Gynecology and Obstetrics, University of Bern, were reviewed. Inclusion criteria was surgically confirmed endometriosis recurrence, defined as a subsequent surgery for endometriosis after a previous complete surgical excision of endometriosis lesions. Three subtypes of endometriosis were defined: superficial peritoneal endometriosis (SUP), ovarian endometrioma (OMA), and deep infiltrating endometriosis (DIE). Time to recurrence and variation in endometriosis subtype between the first and recurrent surgeries were the primary outcome measures. Out of the 322 patients with recurrent surgery that were identified, for 234 of them, the endometriosis subtype at first surgery was confirmed and classified (SUP = 56, OMA = 124, DIE = 54). No statistically significant difference was found for time to recurrence between lesion subtypes. SUP compared to the other groups had a higher possibility of presenting with SUP at recurrence (Odds Ratio (OR): 3.65, 95% confidence interval (CI): 1.74–7.51) and OMA compared to the other groups had a higher possibility of presenting with OMA at recurrence (OR: 3.72, 95% CI: 2.04–6.74). Nevertheless, a large number of SUP patients subsequently presented with OMA (10/56: 17.9%) or DIE (27/56: 48.2%) lesions at recurrence. Similarly, a large number of OMA patients subsequently presented with DIE (49/124: 39.5%) lesions at recurrence. In conclusion, although SUP and OMA patients compared to the others are more likely to present with the same subtype at recurrence, increasing lesion subtype severity occurs in a substantial proportion of patients. Time to recurrence is independent from the lesion subtype at first surgery.