Scientific Abstract Presentations: Research & Technology 9 Sunday, October 29 Sun, Oct 29 4:00-5:00 p.m. CDT

Introduction Post-mastectomy implant infections range from 2- 28%,¹⁻² occurring much more frequently than the 1-2% infection rate reported in cosmetic breast augmentation.³ With the evolution of microbiome science, we can now examine and define the unique composition of microorganisms in one's body and how it may play a role in a person's health outcomes. Previous studies have suggested a significant difference in the breast microbiome between cancer and non-cancer patients,⁴⁻⁵ and this may have the potential to influence infection rates. We present our pilot study using 16s rRNA sequencing to characterize the breast microbiome in mastectomy patients both intraoperatively and post-operatively, comparing cancer patients to those undergoing prophylactic risk-reducing mastectomy.

Methods A prospective randomized-controlled trial was designed for mastectomy patients undergoing two-stage implant-based breast reconstruction. Intraoperatively, a 1cm breast tissue specimen was collected, and post-operatively the peri-prosthetic space was sampled via expander aspiration or drain output at two time points (1-2 weeks and 3-4 weeks). Microbial analysis was performed with 16S rRNA microbiome sequencing. The top represented species and relative abundance percentage of various microbial signals in each sample were recorded.

Results Of the 37 enrolled patients with intra-operative breast tissue and post-operative aspirate samples, 23 (62%) patients had invasive cancer, 6 (16%) had carcinoma in situ, and 8 (22%) patients underwent prophylactic risk-reducing mastectomies. The most represented genus in the breast at time of surgery varied significantly between cancer patients, in situ patients, and non-cancer patients(p=0.045), but the peri-prosthetic breast aspirates did not post-operatively (p=0.593). Pseudomonas was the top species in 50% of prophylactic tissue samples, compared to 43% of invasive cancer patients, and 33% of in-situ patients. However post-operatively, Pseudomonas was the top represented species in only 12.5% of prophylactic samples, compared to 43.5% and 66.7% of carcinoma and carcinoma in situ patients respectively. When comparing invasive carcinoma and carcinoma in situ patients to non-cancer patients, the mean relative abundance percentage of each signal in the sample showed no difference for Pseudomonas, Staphyloccocus, Cornyebacterium, Bradyrhizobium and Streptococcus (p>0.05). However, in surgery there was a significantly lower abundance of Acinetobacter in non-cancer patient tissue (1.1% vs 6.0%, p=0.005), and Burkholderia (0.5% vs 3.9%, p=0.037) compared to cancer patients. This changed post-operatively when there was higher Acinetobacter in non-cancer patients at 1-2 weeks (9.2% vs 2.5%, p=0.023).

Conclusion We present the first study to look at the local breast microbiome at time of mastectomy and post-operatively. After surgery, there were different species represented as the top species between prophylactic patients and cancer patients, which may reflect cancer-related immune changes to the balance of the breast microbiome in response to stress and surgery. Further studies are critical to understanding the implications of these differences and how to potentially optimize the balance of microorganisms for improved outcomes.

References 1. Frey JD, Choi M, Salibian AA, Karp NS. Comparison of Outcomes with Tissue Expander, Immediate Implant, and Autologous Breast Reconstruction in Greater Than 1000 Nipple-Sparing Mastectomies. Plast Reconstr Surg. Jun 2017;139(6):1300-1310. doi:10.1097/PRS.0000000000003340 2. Poppler LH, Mundschenk MB, Linkugel A, Zubovic E, Dolen UC, Myckatyn TM. Tissue Expander Complications Do Not Preclude a Second Successful Implant-Based Breast Reconstruction. Plast Reconstr Surg. 01 2019;143(1):24-34. doi:10.1097/PRS.0000000000005131 3. Pittet B, Montandon D, Pittet D. Infection in breast implants. Lancet Infect Dis. Feb 2005;5(2):94-106. doi:10.1016/S1473-3099(05)01281-8 4. Campbell MJ, McCune E, Johnson B, et al. Breast cancer and the human oral and gut microbiomes [abstract]. Proceedings of the American Association for Cancer Research Annual Meeting 2019; Atlanta, GA Philadelphia (PA): AACR; . Cancer Res 2019;79(13 Suppl):Abstract nr 2830.2019. 5. Tzeng A, Sangwan N, Jia M, et al. Human breast microbiome correlates with prognostic features and immunological signatures in breast cancer. Genome Med. Apr 16 2021;13(1):60. doi:10.1186/s13073-021-00874-2

Purpose As artificial intelligence (AI) innovation blossoms, minimal advancements have occurred in its integration within plastic surgery; this is especially disadvantageous given AI's potential for improved patient outcomes and experience. Most recently, a novel machine learning (ML) model using Generative Pre-Trained Transformers (GPT) has been making headlines for its ability to pass the United States Medical Licensing Exam (USMLE) and its capability to converse with the public on a wide array of topics. Chat GPT3 developed by OpenAI and released in late 2022 utilizes a deep learning model through neural networks to recognize data patterns, and through supervised and reinforced human learning is able to answer a broad range of questions. With the US healthcare system facing a physician shortage, increasingly shorter clinical visits, and a substantial administrative burden, we investigated Chat GPT's capability and accuracy in addressing common peri-operative questions by plastic surgery patients, with the ultimate goal of utilizing a GPT model as an adjunct to assist surgeons in peri-operative care.

Methods Misconceptions on various common plastic surgery procedures and plastic surgery as a field among the public were identified using a literature search. Surveys on breast reconstruction, silicone implants, bariatric surgery, and preconceptions of cosmetic versus plastic surgery were adapted into questions for the Chat GPT platform (1-4). Chat GPT answers were then assessed for accuracy and compared to published literature addressing these misconceptions.

Results In addressing questions on common misconceptions regarding various plastic surgery procedures including risks and complications, Chat GPT answered 100% of the questions correctly. However, when answering questions on PRS procedure costs Chat GPT accuracy dropped to 30% and was at the lower range of price estimated when compared to ASPS. Lastly, in addressing differences between plastic and cosmetic surgery, it answered 62.8% of questions correctly, and frequently confused the term plastic and cosmetic surgeons, which may lead to further public confusion. The model did have a preference towards plastic surgeons relative to other providers and surgical sub-specialties when asked to decide between subspecialties for common plastic surgery procedures such as breast implants and rhinoplasty.

Conclusion
ChatGPT's ability to answer common perioperative questions and misconceptions with 100% accuracy illustrates its broad medical knowledge. While the model was able to answer most questions accurately, its answers were often basic and not nuanced. Regardless, a GPT AI model has significant potential to be a clinical adjunct to aid patients in answering peri-operative questions to allow for more enhanced and efficient patient care.

1. Gusenoff JA Pennino RP, Messing S et al. Post-Bariatric Surgery Reconstruction: Patient Myths, Perceptions, Cost, and Attainability Strategies. Plastic and Reconstructive Surgery. 2008;122(1)
2. Schneider LF, Mehrara BJ. De-Mythifying Breast Reconstruction: A Review of Common Misconceptions about Breast Reconstruction. Journal of the American College of Surgeons. 2015;220(3)
3. Shah A, Patel A, Smetona J, Rohrich RJ. Public Perception of Cosmetic Surgeons versus Plastic Surgeons: Increasing Transparency to Educate Patients. Plastic and Reconstructive Surgery. 2017;139(2)
4. Rohrich RJ, Kaplan J, Dayan E. Silicone Implant Illness: Science versus Myth? Plastic and Reconstructive Surgery. 2019;144(1)

Purpose A groundswell of research demonstrates that emotions impact memory, decision-making attention, and risk tolerance. However, the psychological management of emotions (i.e., emotional regulation [ER]) is missing from surgical education discourse and surgical training curricula. Although emotions were long thought to be a hindrance to cognition, literature demonstrates that they are not easily suppressed; indeed, suppressed emotions come at the cost of individual self-esteem, growth, and environmental mastery.1

Conversely, stress, is frequently found within surgical literature.2–4 One study found that 80% of surgeons surveyed felt the need for stress management training.3 Further, stress has been shown to decrease technical and non-technical skills, both in the operating room and simulations.2,4 Mental skills training (MST) has been shown to help professionals attain higher levels of mental and technical performance. MST includes cognitive and behavioral techniques that focus on learning about the effects of acute stress on performance, acquiring and rehearsing coping strategies to optimize performance, and applying these strategies to real-world, stressful situations.5 However, MST does not incorporate ER despite the potential for emotions' impact on surgeon performance.

We aimed to review the presence and effectiveness of ER and MST in postgraduate surgical training programs. Future research will be conducted to survey North American plastic and reconstructive program directors regarding the current use of ER or MST training in surgical curricula.

Methods A systematic search was conducted on Medline, Embase, and APA PsychInfo from January 1 2000 to February 3 2023 following PRISMA guidelines. MeSH terms and keywords included"surgical education," "emotional regulation," and "mental skills training." Articles were selected based on predetermined eligibility criteria.

Results The initial search yielded 1528 articles. After duplicate removal, 1345 articles were screened by title and abstract. A total of 89 articles proceeded to full-text review; of those, 50 articles were included. The majority of studies were conducted amongst general surgery residents (28% of studies), undergraduate medical students (20%), or surgical residents sub-speciality unspecified. MST was found to be positively associated with decreased anxiety (n=8 studies), improved technical skills (n=14), and improved problem solving ability (n=6). Only one study evaluated the integration of MST within formal curricula, while no studies mentioned the use of ER training. Many studies evaluated the benefit of ER and MST in the context of surgical simulations, only one of which evaluated ER training within a plastic and reconstructive surgery (PRS) program.

Conclusions MST is an effective method to manage stress and improve performance amongst surgical residents. While this is a notable advance in improving surgeon performance and wellbeing, it is yet to be adopted throughout PRS training programs in North America. Further, the use of ER remains absent from MST and formal surgical education leaving a further missed opportunity to improve performance and wellbeing. The results of this systematic review represent a static picture of an evolving landscape. Future research is needed to directly survey surgical educators regarding the formal incorporation of MST and ER within surgical programs.

[1] LeBlanc VR, McConnell MM, Monteiro SD. Predictable chaos: a review of the effects of emotions on attention, memory and decision making. Adv Health Sci Educ Theory Pract. 2015;20(1):265-282. doi:10.1007/s10459-014-9516-6 [2] Bajunaid K, Mullah MAS, Winkler-Schwartz A, Alotaibi FE, Fares J, Baggiani M, et al. Impact of acute stress on psychomotor bimanual performance during a simulated tumor resection task. J Neurosurg. 2017;126(1):71-80. doi:10.3171/2015.5.JNS15558 [3] Anton NE, Montero PN, Howley LD, Brown C, Stefanidis D. What stress coping strategies are surgeons relying upon during surgery? Am J Surg. 2015;210(5):846-851. doi:10.1016/j.amjsurg.2015.04.002 [4] Anton NE, Athanasiadis DI, Karipidis T, Keen AY, Karim A, Cha J, et al. Surgeon stress negatively affects their non-technical skills in the operating room. Am J Surg. 2021;222(6):1154-1157. doi:10.1016/j.amjsurg.2021.01.035 [5] McDonald J, Orlick T, Letts M. Mental readiness in surgeons and its links to performance excellence in surgery. J Pediatr Orthop. 1995;15(5):691-697. doi:10.1097/01241398-199509000-00027

Background: The mouse tail model is frequently used to study post-surgical lymphedema. We have anecdotally observed that male mice have significantly more inflammation and swelling compared with female mice. Although primary lymphedema is more common in females, the effect of sex on secondary lymphedema remains largely unknown. The purpose of this study was, therefore, to study the effect of sex on tail lymphedema in the mouse tail model. Because females are protected from reactive oxygen injury and research studies suggest that lymphatic injury can increase the concentration of reactive oxygen and reactive nitrogen species (ROS and RNS, respectively), we also tested the hypothesis that these changes also contribute to increased inflammation and swelling in male mice.

Methods: We performed microsurgical tail lymphatic excision on male and female wildtype (WT) and constitutive iNOS knockout (iNOS-KO) mice. To assess the progression of the lymphedema phenotype, we performed weekly tail diameter measurements over a 6-week period. Each week, we also recorded the number of necrosed tails in each group. To analyze histological changes associated with lymphedema progression, we performed H&E staining to measure fibroadipose thickness. Additionally, we assessed immune cell tissue infiltration using immunohistochemistry and flow cytometry with anti-CD45 antibodies.

Results: WT male mice had markedly increased tail swelling shortly after surgery compared with female mice. Using Kaplan-Meier survival analysis, we found higher rates of tail necrosis (due to extreme swelling) in WT male mice compared to females. In addition, WT male mice that did not have tail necrosis tended to have increased swelling at each time point after surgery and this effect was most notable 6 weeks postop. Interestingly, we found that iNOS knockout males had decreased rates of necrosis compared with WT mice; although tail volumes did not significantly differ in the tails that did survive until 6 weeks postoperatively. The loss of iNOS in female mice had no effect on tail necrosis or tail volumes over the course of the experiment. Histological analysis revealed that iNOS KO male mice had decreased dermal fibroadipose deposition at 6 weeks postop compared with WT males. These changes correlated with decreased numbers of CD45+ (pan-leukocyte marker) cells on immunohistochemistry and flow cytometry in male iNOS-KO mice versus male WT controls.

Conclusions: Our preliminary findings suggest that male mice have an increased propensity for developing inflammation and oxidative stress after lymphatic injury. It is possible, therefore, that hormonal agents used for treatment of breast cancer may have an effect on the development of lymphedema. Our findings further suggest that anti-oxidative treatments may have some efficacy for preventing/treating lymphedema. Finally, our findings suggest that other mechanisms may be responsible for the increased rates of primary lymphedema in females. Future studies will determine how oxidative stress injures lymphatics, and how sex-related differences contribute to primary or secondary lymphedema.

Purpose 

Our group has developed a custom digital application to assess objective hand function using an Apple pen and iPad to extract geometric drawing features from specific drawing modules. We performed an initial validation study of this novel technique by assessing the ability: (1) To differentiate patients from controls for both dominant and non-dominant hands, and (2) To assess the correlation of geometric drawing features with previously validated patient-reported outcome scores of upper extremity and global function. 

Methods 

This is a prospective study of patients with both hand-wrist and non-hand-wrist pathologies. Participants were asked to draw multiple shapes on an Apple iPad with a digital pen. The drawings from 142 hands in 73 participants were categorized into four groups (dominant/non-dominant hand and patient/control). The raw data collected by the app included pen coordinates, pressure, azimuth, and altitude over time. We calculated kinematic and pressure-based features that generalize to any drawn shape from the raw data. Machine learning models were then used to statistically classify patients and controls, and to create composite scores. Model performance for classification was assessed using accuracy, precision, recall, F1 score, and area under the curve (AUC). Model performance for predicting composite scores was assessed using absolute error. 

Results 

Patients and controls could not be differentiated by simple visual inspection of drawings; however, many geometric features were significantly different (p<0.01) between patients and controls for both dominant and non-dominant hand drawings.  The circle drawings were the most informative and pressure features were the most important. The dominant and non-dominant hand classification metrics for discriminating patients from controls were similar (AUC = ~ 0.85, Accuracy = ~0.75, F1 = ~0.80). Composite geometric drawing features were significantly correlated (p < 0.001) with PRWE, SF12, and qDASH scores.

Conclusion 

• We developed a novel technique to objectively measure hand function using a drawing app
•  Geometric drawing features could differentiate patients with hand pathologies from controls without hand pathologies, regardless of hand dominance  
• Geometric drawing features are correlated with validated patient-reported outcomes scores. 

Introduction: Abstract screening in systematic reviews requires expertise and a considerable amount of time. Machine learning models can learn from examples during the abstract screening and expedite the process. The study aimed to validate a machine learning model in plastic surgery systematic reviews.

Methods: We used the abstracts of two recent plastic surgery systematic reviews that were completely screened by at least two reviewers to build the structure of the model. Then, we applied the model on a new systematic review. We built a recurrent neural network with long short-term memory and tuned the hyperparameters of the network based on the validation subset. We randomly split the data into two equal parts. The first half was further divided into a training subset (60%) and a validation subset (40%), while the second part was used for testing the model for abstracts that were never used in the model training or tuning. The model screened the second half of the abstracts in less than 1 minute. We compared the model predictions with two independent reviewers that were blinded to the model predictions. All the conflicts were resolved by a human reviewer, and all the reasons for mispredictions were explored.

Results: The prospective systematic review had 4628 abstracts. The receiver operatic characteristic curve had an area under the curve of 96%. The sensitivity and specificity of the model were 75% and 98%, respectively. The accuracy was 97%. The model improved the efficiency of abstract screening by 25% (30 hours). The model was used as a validity check to re-evaluate the misclassified abstracts between the model and the reviewers to improve the robustness of the results. The model was correct in 82% of all the conflicts. The most common reason for misprediction by the model was the identification of published protocols from eligible studies.

Conclusion: Deep learning using recurrent neural networks can improve the efficiency and robustness of abstract screening in plastic surgery systematic reviews. Recurrent neural networks are versatile and can be used to improve efficiency, combined with traditional abstract screening to ensure accuracy, and promptly screen new abstracts.

Purpose: The gold standard for diagnosis of acute rejection (AR) in vascularized composite allotransplantation (VCA) is skin biopsy followed by dermatohistopathologic evaluation and grading via the Banff criteria. Informed largely by experience in solid organ transplantation, the Banff criteria emphasizes inflammatory infiltrate, as well as epithelial and adnexal involvement. However, recently there has been growing concern regarding the validity of the Banff criteria for AR diagnosis in VCA (1). In this study, we trained a machine learning model called a convolutional neural network (CNN) to classify images of VCA skin biopsies as either rejecting or not-rejecting. After which, class activation mapping (CAM) was employed to generate heatmaps highlighting areas of interest identified by the CNN. These areas were assessed for congruency with areas of interest identified by the Banff criteria.

Methods: Digital skin biopsy slides from face transplant recipients were sourced. Images of multiple non-overlapping segments were taken from each slide. These Images were tagged as rejecting or not-rejecting based on Banff grade and composite clinical diagnosis. This data was used to train a CNN, with a train/dev/test split of 70/20/10. CAM was utilized to visualize the relative importance of each pixel in the images. Two reviewers independently examined the CAM overlay images and identified both the total number of areas of focus emphasized by the CNN as well as which of the areas of focus coincided with regions of importance according to the Banff criteria.

Results: A total of 307 images, 179 not-rejecting and 130 rejecting, were taken from 57 biopsies in three face transplant recipients. Following model development, 172 (96.1%) not-rejecting images and 120 (92.3%) rejecting images were correctly identified for a precision of 94.5% and a recall of 94.5%. Mean model confidence on rejecting and non-rejecting images correctly identified was 93.7% and 97.0%, respectively. Alternatively, on rejecting images that were incorrectly classified as not-rejecting, mean confidence was 88.5%, while on not rejecting images that were classified as rejecting, confidence was 68.9%. Following the evaluation of the CAM heatmaps, the two reviewers found 771 areas of interest across 307 slides (2.51 area/slide). Of these areas, 470 (61.0%) overlapped with Banff Criteria identified regions, with epithelia being more commonly highlighted than adnexal or perivascular structures.

Conclusion: This study demonstrated a highly accurate CNN could identify rejection in VCA recipients despite emphasizing regions largely different than those identified by the Banff criteria. While this model is not for clinical translation, it identifies considerable discordance between regions highlighted by this CNN and those emphasized by the Banff criteria. These findings contribute to the growing body of evidence highlighting the potential limitations of Banff grading as the sole approach to AR diagnosis in VCA, and the importance of a more holistic, and multimodal approach to AR diagnosis in these patients.   References: (1) Schneider M, Cardones AR, Selim MA, Cendales LC. Vascularized composite allotransplantation: a closer look at the banff working classification. Transpl Int. 2016;29(6):663-671. doi:10.1111/tri.12750

Introduction: Microsurgery is a highly technical and resource-intensive subspecialty within the field of plastic surgery that is often not accessible in many hospital systems. This lack of accessibility is particularly prevalent in foreign nations which do not have access to expert microsurgical training. Surgeons and trainees around the world often have access to microscopes but lack the microsurgical guidance to complete the complex maneuvers associated with operations such as free flaps and peripheral nerve repairs. Here, we propose a highly accessible and low-cost method to livestream microsurgical education over popular online platforms in order to lift the current barriers associated with microsurgical education. Methods: A three camera system was developed to provide a complete and seamless view of the microsurgical field, instruments, and microsurgeon to livestream on any platform. These included a camera tethered to a cam link for a direct view of the microsurgical field, a second camera to view the microsurgeon's hands and instruments, and a third camera to view the microsurgeon's face. A microphone was also placed near the microsurgeon to enable clear audio during the operation. Open Broadcasting Software was used to compile this system onto one page to share over any streaming platform. Results: Six microsurgical livestreams were completed at the University of Wisconsin-Madison over Zoom and Instagram platforms. These events were shared one day prior on Facebook and Instagram. A total of 96 surgeons and trainees tuned into the livestreams representing 28 countries worldwide.
Conclusions: Microsurgery education is a highly complex and specialized field within plastic surgery that is often overlooked due to lack of proper equipment and training opportunities. Here, we propose an accessible and low-cost method to deliver virtual microsurgery education in order to overcome many of the educational barriers associated with this field.