Robust deep learning-based protein sequence design using ProteinMPNN

Although deep learning has revolutionized protein structure prediction, almost all experimentally characterized de novo protein designs have been generated using physically based approaches such as Rosetta. Here, we describe a deep learning-based protein sequence design method, ProteinMPNN, that has outstanding performance in both in silico and experimental tests. On native protein backbones, ProteinMPNN has a sequence recovery of 52.4% compared with 32.9% for Rosetta. The amino acid sequence at different positions can be coupled between single or multiple chains, enabling application to a wide range of current protein design challenges. We demonstrate the broad utility and high accuracy of ProteinMPNN using x-ray crystallography, cryo-electron microscopy, and functional studies by rescuing previously failed designs, which were made using Rosetta or AlphaFold, of protein monomers, cyclic homo-oligomers, tetrahedral nanoparticles, and target-binding proteins.

Protocol for a randomized controlled multicenter trial assessing the efficacy of leuprorelin for severe polycystic liver disease: the AGAINST-PLD study

Background

In patients with severe polycystic liver disease (PLD), there is a need for new treatments. Estrogens and possibly other female sex hormones stimulate growth in PLD. In some patients, liver volume decreases after menopause. Female sex hormones could therefore be a target for therapy. The AGAINST-PLD study will examine the efficacy of the GnRH agonist leuprorelin, which blocks the production of estrogen and other sex hormones, to reduce liver growth in PLD.

Methods

The AGAINST-PLD study is an investigator-driven, multicenter, randomized controlled trial. Institutional review board (IRB) approval was received at the University Medical Center of Groningen and will be collected in other sites before opening these sites. Thirty-six female, pre-menopausal patients, with a very large liver volume for age (upper 10% of the PLD population) and ongoing liver growth despite current treatment options will be randomized to direct start of leuprorelin or to 18 months standard of care and delayed start of leuprorelin. Leuprorelin is given as 3.75 mg subcutaneously (s.c.) monthly for the first 3 months followed by 3-monthly depots of 11.25 mg s.c. The trial duration is 36 months. MRI scans to measure liver volume will be performed at screening, 6 months, 18 months, 24 months and 36 months. In addition, blood will be drawn, DEXA-scans will be performed and questionnaires will be collected. This design enables comparison between patients on study treatment and standard of care (first 18 months) and within patients before and during treatment (whole trial). Main outcome is annualized liver growth rate compared between standard of care and study treatment. Secondary outcomes are PLD disease severity, change in liver growth within individuals and (serious) adverse events. The study is designed as a prospective open-label study with blinded endpoint assessment (PROBE).

Discussion

In this trial, we combined the expertise of hepatologist, nephrologists and gynecologists to study the effect of leuprorelin on liver growth in PLD. In this way, we hope to stop liver growth, reduce symptoms and reduce the need for liver transplantation in severe PLD.

Trial registration Eudra CT number 2020-005949-16, registered at 15 Dec 2020. https://www.clinicaltrialsregister.eu/ctr-search/search?query=2020-005949-16.

Machine learning in the differentiation of follicular lymphoma from diffuse large B-cell lymphoma with radiomic [18F]FDG PET/CT features

BACKGROUND

One of the challenges in the management of patients with follicular lymphoma (FL) is the identification of individuals with histological transformation, most commonly into diffuse large B-cell lymphoma (DLBCL). [¹⁸F]FDG-PET/CT is used for staging of patients with lymphoma, but visual interpretation cannot reliably discern FL from DLBCL. This study evaluated whether radiomic features extracted from clinical baseline [¹⁸F]FDG PET/CT and analyzed by machine learning algorithms may help discriminate FL from DLBCL.

MATERIALS AND METHODS

Patients were selected based on confirmed histopathological diagnosis of primary FL (n=44) or DLBCL (n=76) and available [¹⁸F]FDG PET/CT with EARL reconstruction parameters within 6 months of diagnosis. Radiomic features were extracted from the volume of interest on co-registered [¹⁸F]FDG PET and CT images. Analysis of selected radiomic features was performed with machine learning classifiers based on logistic regression and tree-based ensemble classifiers (AdaBoosting, Gradient Boosting, and XG Boosting). The performance of radiomic features was compared with a SUV_max-based logistic regression model.

RESULTS

From the segmented lesions, 121 FL and 227 DLBCL lesions were included for radiomic feature extraction. In total, 79 radiomic features were extracted from the SUVmap, 51 from CT, and 6 shape features. Machine learning classifier Gradient Boosting achieved the best discrimination performance using 136 radiomic features (AUC of 0.86 and accuracy of 80%). SUV_max-based logistic regression model achieved an AUC of 0.79 and an accuracy of 70%. Gradient Boosting classifier had a significantly greater AUC and accuracy compared to the SUV_max-based logistic regression (p≤0.01).

CONCLUSION

Machine learning analysis of radiomic features may be of diagnostic value for discriminating FL from DLBCL tumor lesions, beyond that of the SUV_max alone.

Multicentre study of multidisciplinary team assessment of pancreatic cancer resectability and treatment allocation

BACKGROUND

Multidisciplinary team (MDT) meetings have been adopted widely to ensure optimal treatment for patients with cancer. Agreements in tumour staging, resectability assessments and treatment allocation between different MDTs were assessed.

METHODS

Of all patients referred to one hospital, 19 patients considered to have non-metastatic pancreatic cancer for evaluation were selected randomly for a multicentre study of MDT decisions in seven units across Northern Europe. Anonymized clinical information and radiological images were disseminated to the MDTs. All patients were reviewed by the MDTs for radiological T, N and M category, resectability assessment and treatment allocation. Each MDT was blinded to the decisions of other teams. Agreements were expressed as raw percentages and Krippendorff's α values, both with 95 per cent confidence intervals.

RESULTS

A total of 132 evaluations in 19 patients were carried out by the seven MDTs (1 evaluation was excluded owing to technical problems). The level of agreement for T, N and M categories ranged from moderate to near perfect (46·8, 61·1 and 82·8 per cent respectively), but there was substantial variation in assessment of resectability; seven patients were considered to be resectable by one MDT but unresectable by another. The MDTs all agreed on either a curative or palliative strategy in less than half of the patients (9 of 19). Only fair agreement in treatment allocation was observed (Krippendorff's α 0·31, 95 per cent c.i. 0·16 to 0·45). There was a high level of agreement in treatment allocation where resectability assessments were concordant.

CONCLUSION

Considerable disparities in MDT evaluations of patients with pancreatic cancer exist, including substantial variation in resectability assessments.

Repeat hepatectomy for recurrent colorectal metastases

BACKGROUND

The oncological benefit of repeat hepatectomy for patients with recurrent colorectal metastases is not yet proven. This study assessed the value of repeat hepatectomy for these patients within current multidisciplinary treatment.

METHODS

Consecutive patients treated by repeat hepatectomy for colorectal metastases between January 1990 and January 2010 were included. Patients undergoing two-stage hepatectomy were excluded. Postoperative outcome was analysed and compared with that of patients who had only a single hepatectomy.

RESULTS

A total of 1036 patients underwent 1454 hepatectomies for colorectal metastases. Of these, 288 patients had 362 repeat hepatectomies for recurrent metastases. Some 225 patients (78·1 per cent) had two hepatectomies, 52 (18·1 per cent) had three hepatectomies, and 11 patients (3·8 per cent) had a fourth hepatectomy. Postoperative morbidity following repeat hepatectomy was similar to that after initial liver resection (27·1 per cent after first, 34·4 per cent after second and 33·3 per cent after third hepatectomy) (P = 0·069). The postoperative mortality rate was 3·1 per cent after repeat hepatectomy versus 1·6 per cent after first hepatectomy. Three- and 5-year overall survival rates following first hepatectomy in patients who underwent repeat hepatectomy were 76 and 54 per cent respectively, compared with 58 and 45 per cent in patients who had only one hepatectomy (P = 0·003). In multivariable analysis, repeat hepatectomy performed between 2000 and 2010 was the sole independent factor associated with longer overall survival.

CONCLUSION

Repeat hepatectomy for recurrent colorectal metastases offers long-term survival in selected patients.

Impact of bevacizumab on functional recovery and histology of the liver after resection of colorectal metastases

BACKGROUND

The impact of bevacizumab on functional recovery and histology of the liver was evaluated in patients undergoing hepatic resection for colorectal liver metastases (CLM) following bevacizumab treatment.

METHODS

Consecutive patients who had resection of CLM between July 2005 and July 2009 following preoperative chemotherapy were identified retrospectively from a prospectively collected database. Patients who had received bevacizumab before the last chemotherapy line were excluded. Postoperative liver function and histology were compared between patients with and without bevacizumab treatment. Recorded parameters included serum prothrombin time, total bilirubin concentration, and levels of aspartate and alanine aminotransferase and γ-glutamyltransferase.

RESULTS

Of 208 patients identified, 67 had received last-line bevacizumab, 44 were excluded and 97 had not received bevacizumab. Most patients in the bevacizumab group (66 per cent) received a single line of chemotherapy. Bevacizumab was most often combined with 5-flurouracil/leucovorin and irinotecan (68 per cent). The median number of bevacizumab cycles was 8·6 (range 1-34). Bevacizumab administration was stopped a median of 8 (range 3-19) weeks before surgery. There were no deaths. Postoperative morbidity occurred in 43 and 36 per cent of patients in the bevacizumab and no-bevacizumab groups respectively (P = 0·353). The mean(s.d.) degree of tumour necrosis was significantly higher in the bevacizumab group (55(27) versus 32(29) per cent; P = 0·001). Complete pathological response rates were comparable (3 versus 8 per cent; P = 0·307). Postoperative changes in functional parameters and objective signs of hepatic toxicity were similar in both groups.

CONCLUSION

Preoperative administration of bevacizumab does not seem to affect functional recovery of the liver after resection of CLM. Tumour necrosis is increased following bevacizumab treatment.

Comparison of simultaneous or delayed liver surgery for limited synchronous colorectal metastases

BACKGROUND

The optimal surgical strategy for patients with synchronous colorectal liver metastases (CLMs) is still unclear. The aim of this study was to compare simultaneous colorectal and hepatic resection with a delayed strategy in patients who had a limited hepatectomy (fewer than three segments).

METHODS

All patients with synchronous CLMs who underwent limited hepatectomy between 1990 and 2006 were included retrospectively. Short-term outcome, overall and progression-free survival were compared in patients having simultaneous colorectal and hepatic resection and those treated by delayed hepatectomy.

RESULTS

Of 228 patients undergoing hepatectomy for synchronous CLMs, 55 (24.1 per cent) had a simultaneous colorectal resection and 173 (75.9 per cent) had delayed hepatectomy. The mortality rate following hepatectomy was similar in the two groups (0 versus 0.6 per cent respectively; P = 0.557), but cumulative morbidity was significantly lower in the simultaneous group (11 per cent versus 25.4 per cent in the delayed group; P = 0.015). Three-year overall and progression-free survival rates were 74 and 8 per cent respectively in the simultaneous group, compared with 70.3 and 26.1 per cent in the delayed group (overall survival: P = 0.871; progression-free survival: P = 0.005). Significantly more recurrences were observed in the simultaneous group at 3 years (85 versus 63.6 per cent; P = 0.002); a simultaneous strategy was an independent predictor of recurrence.

CONCLUSION

Combining colorectal resection with a limited hepatectomy is safe in patients with synchronous CLMs and associated with less cumulative morbidity than a delayed procedure. However, the combined strategy has a negative impact on progression-free survival.

Impact of portal vein embolization on long-term survival of patients with primarily unresectable colorectal liver metastases

BACKGROUND

: Portal vein embolization (PVE) increases the resectability of initially unresectable colorectal liver metastases (CLM). This study evaluated long-term survival in patients with CLM who underwent hepatectomy following PVE.

METHODS

: In a retrospective analysis patients treated by PVE before major hepatectomy were compared with those who did not have PVE, and with those who had PVE without resection.

RESULTS

: Of 364 patients who underwent hepatectomy, 67 had PVE beforehand and 297 did not. Those who had PVE more often had more than three liver metastases (68 versus 40.9 per cent; P < 0.001) that were more frequently bilobar (78 versus 55.2 per cent; P < 0.001), and a higher proportion underwent extended hepatectomy (63 versus 18.1 per cent; P < 0.001). Postoperative morbidity rates were 55 and 41.1 per cent respectively (P = 0.035), and overall 3-year survival rates were 44 and 61.0 per cent (P = 0.001). Thirty-two other patients who were treated by PVE but did not undergo resection all died within 3 years.

CONCLUSION

: PVE increased the resectability rate of initially unresectable CLM. Among patients who had PVE, long-term survival was better in those who had resection than in those who did not. PVE is of importance in the multimodal treatment of advanced CLM.

Liver resection of colorectal metastases in elderly patients

Background

This study evaluated the outcome of liver surgery for colorectal metastases (CLM) in patients over 70 years old in a large international multicentre cohort.

Methods

Among 7764 patients who had resection of CLM, 999 (12·9 per cent) were aged 70–75 years, 468 (6·0 per cent) were aged 75–80 years and 157 (2·0 per cent) were at least 80 years old. Elderly patients were compared with the younger population.

Results

Multinodular and bilateral metastases were less common in elderly than in younger patients (P &lt; 0·001). Preoperative chemotherapy was used less frequently and more limited surgery was performed (P &lt; 0·001). Sixty-day postoperative mortality and morbidity rates were 3·8 and 32·3 per cent respectively, compared with 1·6 and 28·7 per cent in younger patients (both P &lt; 0·001). Three-year overall survival was 57·1 per cent in elderly and 60·2 per cent in younger patients (P &lt; 0·001), and was similar among patients aged 70–75, 75–80 or at least 80 years (57·8, 55·3 and 54·1 per cent respectively; P = 0·160). Independent predictors of survival were more than three metastases, bilateral metastases, concomitant extrahepatic disease and no postoperative chemotherapy.

Conclusion

Liver resection for CLM in elderly patients can achieve a reasonable 3-year survival rate, with an acceptable morbidity rate. There should be no upper age limit but risk factors may help predict potential benefit.

Long-term outcome in patients with adrenal metastases following resection of colorectal liver metastases

BACKGROUND

The prognostic significance of adrenal metastases (AMs) in patients with colorectal liver metastases (CLMs) remains unknown. The aim of this study was to determine the influence of AMs on long-term outcome and the role of adrenalectomy in patients with CLMs.

METHODS

All patients resected for CLMs who developed AMs at a single institution between 1992 and 2006 were included in the study. Their long-term outcome was compared with that of all other patients resected for CLMs but without AMs.

RESULTS

Hepatectomy was performed in 796 patients, of whom 14 (1.8 per cent) developed AMs, a median of 28 months after initial diagnosis of CLMs; the remaining 782 patients (98.2 per cent) had no AMs. All 14 patients had chemotherapy, and ten went on to adrenalectomy. Median survival after diagnosis of CLMs was 50 months in patients with AMs versus 68 months in those without (P = 0.020). After diagnosis of AMs, median survival was 23 months, whether or not adrenalectomy was performed.

CONCLUSION

The development of AMs after liver resection for colorectal cancer deposits carries a poor prognosis, and adrenalectomy is probably not warranted.

Liver regenerative nodular hyperplasia consecutive to preoperative chemotherapy: Impact on outcome of liver surgery for colorectal metastases

4097

Background: Regenerative nodular hyperplasia (RNH) represents the worst evolutive stage of vascular lesions induced by prolonged chemotherapy on the liver. Its incidence and impact on the outcome of resection for colorectal liver metastases (CLM) are however unknown. We evaluated the effect of RNH on postoperative morbidity and assessed its evolution in time at repeat hepatectomy. Methods: All patients that underwent hepatectomy for CLM between January 1990 and November 2006 after 1 line of chemotherapy of more than 6 cycles were included. Detailed histopathologic analysis of the nontumoral liver was performed at first and repeat hepatectomies according to a standard format. Results: Of 146 included patients, 24 (16%) received 5- fluorouracil (5-FU) and leucovorin (LV) alone, 92 (63%) had 5-FU, LV and oxaliplatin, 18 (12%) had 5-FU, LV and irinotecan, and 12 (8%) were treated by 5-FU, LV, oxaliplatin and irinotecan. Overall, RNH occurred in 22 of 146 patients (15%). Patients treated by oxaliplatin more often had RNH compared to oxaliplatin-naïve patients (22% vs 4%) (P=0.003). The presence of RNH was associated with increased postoperative hepatic morbidity (23% vs 11%) (P=0.05). None of the RNH patients died within 60 days postoperatively. A preoperative gamma-glutamyltransferase level >80 U/L and a total bilirubin level >15 μmol/L were independent predictive factors of RNH at first hepatectomy. Two of 22 patients with RNH underwent repeat hepatectomy. RNH disappeared at second hepatectomy in both patients following prolonged treatment with irinotecan. Conclusions: Patients with CLM that receive prolonged courses of preoperative oxaliplatin have an increased risk of RNH and associated postoperative hepatic morbidity. Upfront treatment with short courses of highly effective regimens may therefore be most appropriate. In addition, the continuation of oxaliplatin should be discouraged in case of RNH to avoid its negative effects on further surgery.

No significant financial relationships to disclose.

Postoperative liver function recovery after hepatic resection for colorectal metastases previously treated with bevacizumab

4093

Background: The influence of bevacizumab on postoperative morbidity in patients with colorectal liver metastases (CLM) submitted to hepatectomy has been evaluated. However, in spite of a potential inhibition of liver regeneration, its impact on postoperative liver function recovery remains unknown. Methods: From July 2005 to December 2007, we evaluated 37 patients treated with bevacizumab prior to hepatectomy. Postoperative evolution of liver function variables was compared with that of 70 bevacizumab-naïve patients. Recorded parameters included prothrombin time, serum aspartate and alanine aminotransferase levels, and serum gamma-glutamyltransferase and total bilirubin levels. Results: Most patients of the bevacizumab group received 1 line of chemotherapy (62% vs 54% in the control group) (P=0.43). Bevacizumab was combined to 5-fluorouracil (5-FU), leucovorin (LV) and irinotecan or 5-FU, LV and oxaliplatin in the last preoperative line, in 57% and 30% of cases, respectively. Patients received a mean total number of 10.4+5.7 and 11.1+6.2 cycles of preoperative chemotherapy in the bevacizumab and control group (P=0.56). The median number of administered bevacizumab cycles was 6 (range: 3–25). Bevacizumab administration was stopped at a median of 7.9 weeks before surgery (range: 3.6–18.0). Major hepatectomies were performed in 41% and 36% of bevacizumab and control patients, respectively (P=0.62). Baseline liver function tests as well as postoperative liver function recovery were similar between patients treated with or without bevacizumab (Table). Postoperative morbidity occurred in 35% of bevacizumab patients and in 43% of control patients (P=0.44). Conclusions: Preoperative bevacizumab treatment has no impact on short-term liver function recovery after hepatic resection for CLM and has no deleterious effect on the incidence of postoperative morbidity.

[Table: see text]

No significant financial relationships to disclose.

[Better treatment for patients with colorectal liver metastases]

Three patients, 61, 58 and 63 years old, presented with non-resectable liver metastases from colorectal cancer. The first patient, a man, who had a solitary lesion in the liver and severe cardiovascular morbidity, was successfully treated with laser-induced interstitial thermotherapy. The second patient, a woman, had large multiple liver metastases and two concomitant isolated pulmonary metastases. Following chemotherapy with fluorouracil, leucovorin and oxaliplatin, all lesions were downsized and a hemihepatectomy and pulmonary wedge resections were able to be performed in two stages. At the last follow-up, both patients were disease-free after 12 and 24 months respectively. The third patient, a man, presenting with multiple synchronous liver metastases, showed a significant decrease of hepatic tumour involvement after six courses of capecitabine. At present he is in a good condition and his disease is stable. Surgical resection ofcolorectal liver metastases leads to a 5-year survival rate of up to 45% in selected patients. Unfortunately, only 10 to 20% of patients are amenable to surgical resection. In the remaining group, a combination of new treatment options using local tumour ablative therapies and novel chemotherapeutic regimens provide alternative strategies with the potential of long-term survival.