Computed tomography-based machine learning for donor lung screening before transplantation

BACKGROUND

Assessment and selection of donor lungs remain largely subjective and experience based. Criteria to accept or decline lungs are poorly standardized and are not compliant with the current donor pool. Using ex vivo computed tomography (CT) images, we investigated the use of a CT-based machine learning algorithm for screening donor lungs before transplantation.

METHODS

Clinical measures and ex situ CT scans were collected from 100 cases as part of a prospective clinical trial. Following procurement, donor lungs were inflated, placed on ice according to routine clinical practice, and imaged using a clinical CT scanner before transplantation while stored in the icebox. We trained and tested a supervised machine learning method called dictionary learning, which uses CT scans and learns specific image patterns and features pertaining to each class for a classification task. The results were evaluated with donor and recipient clinical measures.

RESULTS

Of the 100 lung pairs donated, 70 were considered acceptable for transplantation (based on standard clinical assessment) before CT screening and were consequently implanted. The remaining 30 pairs were screened but not transplanted. Our machine learning algorithm was able to detect pulmonary abnormalities on the CT scans. Among the patients who received donor lungs, our algorithm identified recipients who had extended stays in the intensive care unit and were at 19 times higher risk of developing chronic lung allograft dysfunction within 2 years posttransplant.

CONCLUSIONS

We have created a strategy to ex vivo screen donor lungs using a CT-based machine learning algorithm. As the use of suboptimal donor lungs rises, it is important to have in place objective techniques that will assist physicians in accurately screening donor lungs to identify recipients most at risk of posttransplant complications.

Development and validation of the lung donor (LUNDON) acceptability score for pulmonary transplantation

There is a chronic shortage of donor lungs for pulmonary transplantation due, in part, to low lung utilization rates in the United States. We performed a retrospective cohort study using data from the Scientific Registry of Transplant Recipients database (2006-2019) and developed the lung donor (LUNDON) acceptability score. A total of 83 219 brain-dead donors were included and were randomly divided into derivation (n = 58 314, 70%) and validation (n = 24 905, 30%) cohorts. The overall lung acceptance was 27.3% (n = 22 767). Donor factors associated with the lung acceptance were age, maximum creatinine, ratio of arterial partial pressure of oxygen to fraction of inspired oxygen, mechanism of death by asphyxiation or drowning, history of cigarette use (≥20 pack-years), history of myocardial infarction, chest x-ray appearance, bloodstream infection, and the occurrence of cardiac arrest after brain death. The prediction model had high discriminatory power (C statistic, 0.891; 95% confidence interval, 0.886-0.895) in the validation cohort. We developed a web-based, user-friendly tool (available at https://sites.wustl.edu/lundon) that provides the predicted probability of donor lung acceptance. LUNDON score was also associated with recipient survival in patients with high lung allocation scores. In conclusion, the multivariable LUNDON score uses readily available donor characteristics to reliably predict lung acceptability. Widespread adoption of this model may standardize lung donor evaluation and improve lung utilization rates.

CT-based Machine Learning for Donor Lung Screening Prior to Transplantation

Background

Assessment and selection of donor lungs remains largely subjective and experience based. Criteria to accept or decline lungs are poorly standardized and are not compliant with the current donor pool. Using ex vivo CT images, we investigated the use of a CT-based machine learning algorithm for screening donor lungs prior to transplantation.

Methods

Clinical measures and ex-situ CT scans were collected from 100 cases as part of a prospective clinical trial. Following procurement, donor lungs were inflated, placed on ice according to routine clinical practice, and imaged using a clinical CT scanner prior to transplantation while stored in the icebox. We trained and tested a supervised machine learning method called dictionary learning , which uses CT scans and learns specific image patterns and features pertaining to each class for a classification task. The results were evaluated with donor and recipient clinical measures.

Results

Of the 100 lung pairs donated, 70 were considered acceptable for transplantation (based on standard clinical assessment) prior to CT screening and were consequently implanted. The remaining 30 pairs were screened but not transplanted. Our machine learning algorithm was able to detect pulmonary abnormalities on the CT scans. Among the patients who received donor lungs, our algorithm identified recipients who had extended stays in the ICU and were at 19 times higher risk of developing CLAD within 2 years post-transplant.

Conclusions

We have created a strategy to ex vivo screen donor lungs using a CT-based machine learning algorithm. As the use of suboptimal donor lungs rises, it is important to have in place objective techniques that will assist physicians in accurately screening donor lungs to identify recipients most at risk of post-transplant complications.

A new lung donor score to predict short and long-term survival in lung transplantation

Background

Donor selection criteria are crucial for a successful lung transplant outcome. Our objective was to develop a new donor score to predict short- and long-term survival and validate it with five existing lung donor scores (Oto, Eurotransplant, Minnesota, Maryland-UNOS, Louisville-UNOS).

Methods

All 454 adult lung transplants at our center between 1992–2015 were included to develop a new score. Discriminative ability for all scores was calculated by the area under time-dependent receiver operating characteristic curves (time-dependent AUC) at 30-day, 1, 5 and 10-year survival, and their fit compared with Akaike’s information criterion. For the new score, five pre-selected donor risk factors were derived: age, diabetes mellitus, smoking history, pulmonary infection, PaO₂/FiO₂-ratio, weighed via simplification of a multiple Cox model, and shrinkage used to avoid overfitting. The score sub-weighting resulted in a total of 17 points.

Results

The existing scores showed predictive accuracy better than chance in prediction of survival of 5-year (AUC 0.58–0.60) to 10-year survival (AUC 0.58–0.64). Our new score had better discriminative ability as the existing scores with regard to 1, 5 and 10-year survival (AUC 0.59, 0.64, 0.66, respectively). Additional adjustment for recipient and surgical procedure variables improved the time-dependent AUC’s slightly. For the secondary outcomes primary graft dysfunction and bronchiolitis obliterans syndrome, the new score showed also a good predictive accuracy.

Conclusions

The proposed Zurich Donor Score is simple, well adapted for the current urge of extended donors use, and shows higher discriminative ability compared to preexisting donor scores regarding short- to long-term survival.

Ex vivo lung CT findings may predict the outcome of the early phase after lung transplantation

Purpose

We developed an ex vivo lung CT (EVL-CT) technique that allows us to obtain detailed CT images and morphologically assess the retrieved lung from a donor for transplantation. After we recovered the lung graft from a brain-dead donor, we transported it to our hospital and CT images were obtained ex vivo before lung transplant surgery. The objective of this study was to investigate the correlation between the EVL-CT findings and post-transplant outcome in patients who underwent bilateral lung transplantation (BLT) or single lung transplantation (SLT).

Methods

We retrospectively reviewed the records of 70 patients with available EVL-CT data who underwent BLT (34 cases) or SLT (36 cases) in our hospital between October 2007 and September 2017. The recipients were divided into 2 groups (control group, infiltration group) according to the findings of EVL-CT of the lung graft in BLT and SLT, respectively. Recipients in the control group were transplanted lung grafts without any infiltrates (BLT control group, SLT control group). Recipients in the infiltration group received lung grafts with infiltrates (BLT infiltration group, SLT infiltration group).

Results

The recipients in the BLT infiltration group showed significantly slower recovery from primary graft dysfunction and a longer mechanical ventilation period and ICU stay period than those in the BLT control group. The mechanical ventilation period was significantly longer in the recipients in the SLT infiltration group than those in the SLT control group.

Conclusion

EVL-CT may predict the outcome of the early phase after lung transplantation.

Short- and Long-term Outcomes After Lung Transplantation From Circulatory-Dead Donors: A Single-Center Experience

BACKGROUND

Donation after cardiac death (DCD) to overcome the donor organ shortage is well accepted in the clinical setting, although long-term outcome after DCD lung transplantation (LTx) remains largely unknown.

METHODS

In this retrospective study, DCD LTx recipients (n = 59) were compared with a cohort of donation after brain death (DBD) LTx recipients (n = 331) transplanted between February 2007 and September 2013; follow-up was until January 1, 2016. Short-term (duration of mechanical ventilation, intensive care unit stay, hospital stay, and highest primary graft dysfunction score within 72 hours) and long-term (chronic lung allograft dysfunction-free and overall survival) follow-up were compared over a median follow-up of 50.5 (±3.7) months for DCD and 66.8 (±1.5) months for DBD.

RESULTS

There were no differences between groups with regard to patient characteristics: age (P = 0.78), underlying disease (P = 0.30) and type of type of LTx (P = 0.10), except sex where more males were transplanted with a DCD donor (62.7%) vs (48.3%, P = 0.048). There was no difference in time on mechanical ventilation (P = 0.59), intensive care unit stay (P = 0.74), highest primary graft dysfunction score (P = 0.67) and hospital stay (P = 0.99). Moreover, chronic lung allograft dysfunction-free (P = 0.86) and overall survival (P = 0.15) did not differ between the DBD and DCD groups.

CONCLUSIONS

In our experience, both short- and long-term outcomes in DCD lung recipients are comparable to that of DBD lung recipients. Therefore, DCD LTx can be considered a safe strategy that significantly increased our transplant activity.

Radiological Analysis of Unused Donor Lungs: A Tool to Improve Donor Acceptance for Transplantation?

Despite donor organ shortage, a large proportion of possible donor lungs are declined for transplantation. Criteria for accepting/declining lungs remain controversial because of the lack of adequate tools to aid in decision-making. We collected, air-inflated, and froze a large series of declined/unused donor lungs and subjected these lung specimens to CT examination. Affected target regions were scanned by using micro-CT. Lungs from 28 donors were collected. Two lungs were unused, six were declined for non-allograft-related reasons (collectively denominated nonallograft declines, n = 8), and 20 were declined because of allograft-related reasons. CT scanning demonstrated normal lung parenchyma in only four of eight nonallograft declines, while relatively normal parenchyma was found in 12 of 20 allograft-related declines. CT and micro-CT examinations confirmed the reason for decline in most lungs and revealed unexpected (unknown from clinical files or physical inspection) CT abnormalities in other lungs. CT-based measurements showed a higher mass and density in the lungs with CT alterations compared with lungs without CT abnormalities. CT could aid in the decision-making to accept or decline donor lungs which could lead to an increase in the quantity and quality of lung allografts.