Weakly supervised annotation-free cancer detection and prediction of genotype in routine histopathology

Deep learning is a powerful tool in computational pathology: it can be used for tumor detection and for predicting genetic alterations based on histopathology images alone. Conventionally, tumor detection and prediction of genetic alterations are two separate workflows. Newer methods have combined them, but require complex, manually engineered computational pipelines, restricting reproducibility and robustness. To address these issues, we present a new method for simultaneous tumor detection and prediction of genetic alterations: The Slide-Level Assessment Model (SLAM) uses a single off-the-shelf neural network to predict molecular alterations directly from routine pathology slides without any manual annotations, improving upon previous methods by automatically excluding normal and non-informative tissue regions. SLAM requires only standard programming libraries and is conceptually simpler than previous approaches. We have extensively validated SLAM for clinically relevant tasks using two large multicentric cohorts of colorectal cancer patients, Darmkrebs: Chancen der Verhütung durch Screening (DACHS) from Germany and Yorkshire Cancer Research Bowel Cancer Improvement Programme (YCR-BCIP) from the UK. We show that SLAM yields reliable slide-level classification of tumor presence with an area under the receiver operating curve (AUROC) of 0.980 (confidence interval 0.975, 0.984; n = 2,297 tumor and n = 1,281 normal slides). In addition, SLAM can detect microsatellite instability (MSI)/mismatch repair deficiency (dMMR) or microsatellite stability/mismatch repair proficiency with an AUROC of 0.909 (0.888, 0.929; n = 2,039 patients) and BRAF mutational status with an AUROC of 0.821 (0.786, 0.852; n = 2,075 patients). The improvement with respect to previous methods was validated in a large external testing cohort in which MSI/dMMR status was detected with an AUROC of 0.900 (0.864, 0.931; n = 805 patients). In addition, SLAM provides human-interpretable visualization maps, enabling the analysis of multiplexed network predictions by human experts. In summary, SLAM is a new simple and powerful method for computational pathology that could be applied to multiple disease contexts. © 2021 The Authors. The Journal of Pathology published by John Wiley & Sons, Ltd. on behalf of The Pathological Society of Great Britain and Ireland.     

Characterization of a third form of the human T cell receptor gamma/delta

A subpopulation of the CD3+ peripheral T lymphocytes express the TCR-gamma/delta complex. Three distinct TCR-gamma forms that differ in size and in the ability to form a disulfide bridge with the TCR-delta subunit have been described. In this study we analyze the structural difference between the non-disulfide-linked 55-kD and 40-kD TCR-gamma chains. The 40-kD TCR-gamma form contains a smaller polypeptide backbone and carries less carbohydrate compared with the 55-kD TCR-gamma form. A cDNA clone corresponding to the 40-kD TCR-gamma subunit lacks one copy of the second exon of the constant region that is present in the other TCR-gamma subunit. This exon copy encodes part of the connector region that is located between the constant domain and the membrane spanning region. We show that the number of potential N-linked glycan attachment sites are the same for the two TCR-gamma forms. Since these attachment sites are located in the connector region we conclude that the connector region influences the amount of N-linked carbohydrates added to the core TCR-gamma polypeptide, probably by affecting the conformation of the protein. In contrast to the TCR-beta constant region usage, the TCR-gamma constant regions are unequally expressed. Virtually exclusive usage of disulfide-linked complexes were found in some individuals, while both the disulfide-linked and the 40-kD, non-disulfide-linked TCR-gamma forms were detected in other subjects. The ability to distinguish these TCR-gamma/delta forms now makes it possible to study the mechanisms that govern their selection and to determine if they correspond to functionally distinct isotypes.     

Discrepancy between sonographic and radiographic values after ultrasound-monitored treatment of developmental dysplasia of the hip

Introduction

The purpose of the study was to examine the accordance between the actually used sonographic and radiographic standard values after ultrasound-monitored treatment of developmental dysplasia of the hip (DDH).

Material and methods

One hundred and fifty-three (119 children) ultrasound-monitored treated hips (initial staging according to Graf: type IIc–IV) which attained normal ultrasound findings (type I according to Graf) during treatment and underwent an anteroposterior radiograph of the pelvis at the time of starting walking (mean age 18.6 months) were evaluated retrospectively.

Results

While all hips showed normal sonographic values (Graf type I), 26 (17%) showed mild and 17 (11.1%) severe dysplasia (by measuring the acetabular index) according to the radiographic Toennis classification system, and 29 (19%) showed mild and 48 (31.4%) severe dysplasia according to the Wiberg centre-edge angle.

Conclusions

This data show that the actually used sonographic and radiographic standard values concerning DDH do not correlate appropriately. It must be put up for discussion whether the radiographic standard values might be too strict. Further criteria must be developed to better assess the prognosis of residual dysplasia.     

Arylsulfatase A pseudodeficiency: loss of a polyadenylylation signal and N-glycosylation site.

Metachromatic leukodystrophy is a metabolic disorder caused by the deficiency of arylsulfatase A. Deficiency of this enzyme is also found in apparently healthy individuals, a condition for which the term pseudodeficiency was introduced. The arylsulfatase A (cerebroside-3-sulfate 3-sulfohydrolase; EC 3.1.6.8) (ASA) encoding gene was isolated from an individual homozygous for the ASA pseudodeficiency allele. Sequence analysis revealed two A----G transitions. One changes Arg-350 to serine, which leads to the loss of a utilized N-glycosylation site. This loss explains the smaller size of ASA in ASA pseudodeficiency fibroblasts. The introduction of Ser-350 into normal ASA cDNA does not affect the rate of synthesis, the stability, or the catalytic properties of ASA in stably transfected baby hamster kidney cells. Therefore, the loss of the N-linked oligosaccharide does not contribute to the reduction of ASA activity in ASA pseudodeficiency. The other A----G transition changes the first polyadenylylation signal downstream of the stop codon from AATAAC to AGTAAC. The latter causes a severe deficiency of a 2.1-kilobase (kb) mRNA species. The deficiency of the 2.1-kb RNA species provides an explanation for the diminished synthesis of ASA seen in pseudodeficiency fibroblasts. Amplification of genomic DNA and hybridization with allele-specific oligonucleotides detected both mutations in four unrelated individuals with ASA pseudodeficiency.     

The accuracy of digital templating: a comparison of short-stem total hip arthroplasty and conventional total hip arthroplasty

Purpose

Selection of the correct femoral stem size is crucial in total hip arthroplasty for an uncomplicated implantation and good initial stability. Pre-operative templating has been shown to be a valuable tool in predicting the correct implant size. For short-stem total hip arthroplasty (SHA), which recently is increasingly used, it is unknown if templating can be performed as reliable as conventional total hip arthroplasty (THA).

Methods

A total of 100 hip arthroplasties, 50 with SHA and 50 with THA, were templated by four orthopaedic surgeons each. The surgeons had different levels of professional experience and performed a digital template of the acetabular and femoral component on the pre-operative radiographs. The results were compared with the truly inserted implant size.

Results

For the femoral stems the average percentage of agreement (±1 size) was 89.0 % in SHA and 88.5 % in THA. There was no significant difference among surgeons in the accuracy of templating the correct stem size and no significant difference between templating SHA and THA. For the acetabular component the average percentage of agreement (±1 size) was 75.8 %. However, the more experienced surgeons showed a significant higher accuracy for templating the correct cup size than the less experienced surgeons.

Conclusion

Digital templating of SHA can predict the stem sizes as accurately as conventional THA. Therefore digital templating is also recommendable for SHA, as it helps to predict the implant size prior to surgery and thereby might help to avoid complications.     

Evaluation of tracheobronchial anomalies in children using low-dose multidetector CT: report of a 13-year-old boy with a tracheal bronchus and recurrent pulmonary infections

Tracheobronchial anomalies in children may be associated with recurrent episodes of pulmonary infections and symptoms of recurrent or persistent airway obstruction. Diagnosis by conventional imaging may be difficult. Multidetector computed tomography (MDCT) offers the possibility to generate a virtual three-dimensional bronchoscopy, thus enabling detailed overview of the tracheobronchial system. We report on a 13-year old boy, admitted to hospital after recurrent episodes of bronchial infections. Functional studies showed airway obstruction with no response to bronchodilators. A chest radiograph was normal. Flexible bronchoscopy revealed tracheobroncho malacia of the distal trachea and the right main bronchus. The ostium of an accessory right-sided tracheal bronchus, which could not be entered by the endoscope, was also detected. MDCT using a low-dose protocol was performed on a four-section scanner (Somatom Volume Zoom, Siemens, Erlangen, Germany). A three-dimensional virtual bronchoscopy based on surface rendering was generated, which confirmed moderate narrowing of the trachea and right main bronchus. Furthermore, an accessory and stenotic tracheal bronchus including poststenotic segments, ventilating parts of the right upper lobe, could be clearly visualized. MDCT can be a valuable instrument in the diagnostic pathway of assessing tracheobronchial anomalies in children, including visualization of poststenotic bronchial structures. The use of low-dose protocols provides adequate image quality to perform virtual bronchoscopy, thus reducing administered radiation to a tolerable amount.