Open access image repositories: high-quality data to enable machine learning research.

Originally motivated by the need for research reproducibility and data reuse, large-scale, open access information repositories have become key resources for training and testing of advanced machine learning applications in biomedical and clinical research. To be of value, such repositories must provide large, high-quality data sets, where quality is defined as minimising variance due to data collection protocols and data misrepresentations. Curation is the key to quality. We have constructed a large public access image repository, The Cancer Imaging Archive, dedicated to the promotion of open science to advance the global effort to diagnose and treat cancer. Drawing on this experience and our experience in applying machine learning techniques to the analysis of radiology and pathology image data, we will review the requirements placed on such information repositories by state-of-the-art machine learning applications and how these requirements can be met.

Radiation Therapy Quality Assurance Analysis of Alliance A021501: Preoperative mFOLFIRINOX or mFOLFIRINOX Plus Hypofractionated Radiation Therapy for Borderline Resectable Adenocarcinoma of the Pancreas.

PURPOSE: Alliance A021501 is the first randomized trial to evaluate stereotactic body radiation therapy (SBRT) for borderline resectable pancreatic ductal adenocarcinoma (PDAC) after neoadjuvant chemotherapy. In this post hoc study, we reviewed the quality of radiation therapy (RT) delivered. METHODS AND MATERIALS: SBRT (6.6 Gy × 5) was intended but hypofractionated RT (5 Gy × 5) was permitted if SBRT specifications could not be met. Institutional credentialing through the National Cancer Institute-funded Imaging and Radiation Oncology Core (IROC) was required. Rigorous RT quality assurance (RT QA) was mandated, including pretreatment review by a radiation oncologist. Revisions were required for unacceptable deviations. Additionally, we performed a post hoc RT QA analysis in which contours and plans were reviewed by 3 radiation oncologists and assigned a score (1, 2, or 3) based on adequacy. A score of 1 indicated no deviation, 2 indicated minor deviation, and 3 indicated a major deviation that could be clinically significant. Clinical outcomes were compared by treatment modality and by case score. RESULTS: Forty patients were registered to receive RT (1 planned but not treated) at 27 centers (18 academic and 9 community). Twenty-three centers were appropriately credentialed for moving lung/liver targets and 4 for static head and neck only. Thirty-two of 39 patients (82.1%) were treated with SBRT and 7 (17.9%) with hypofractionated RT. Five cases (13%) required revision before treatment. On post hoc review, 23 patients (59.0%) were noted to have suboptimal contours or plan coverage, 12 (30.8%) were scored a 2, and 11 (28.2%) were scored a 3. There were no apparent differences in failure patterns or surgical outcomes based on treatment technique or post hoc case score. Details related to on-treatment imaging were not recorded. CONCLUSIONS: Despite rigorous QA, we encountered variability in simulation, contouring, plan coverage, and dose on trial. Although clinical outcomes did not appear to have been affected, findings from this analysis serve to inform subsequent PDAC SBRT trial designs and QA requirements.

Local Control For High-Grade Nonrhabdomyosarcoma Soft Tissue Sarcoma Assigned to Radiation Therapy on ARST0332: A Report From the Childrens Oncology Group.

PURPOSE: The ARST0332 trial for pediatric and young adults with nonrhabdomyosarcoma soft tissue sarcoma (NRSTS) used risk-based treatment including primary resection with lower-than-standard radiation doses to optimize local control (LC) while minimizing long-term toxicity in those requiring radiation therapy (RT). RT for high-grade NRSTS was based on extent of resection (R0: negative margins, R1: microscopic margins, R2/U: gross disease/unresectable); those with >5 cm tumors received chemotherapy (CT; ifosfamide/doxorubicin). This analysis evaluates LC for patients assigned to RT and prognostic factors associated with local recurrence (LR). METHODS AND MATERIALS: Patients aged <30 years with high-grade NRSTS received RT (55.8 Gy) for R1 ≤5 cm tumor (arm B); RT (55.8 Gy)/CT for R0/R1 >5 cm tumor (arm C); or neoadjuvant RT (45 Gy)/CT plus delayed surgery, CT, and postoperative boost to 10.8 Gy R0 <5 mm margins/R1 or 19.8 Gy for R2/unresected tumors (arm D). RESULTS: One hundred ninety-three eligible patients had 24 LRs (arm B 1/15 [6.7%], arm C 7/65 [10.8%], arm D 16/113 [14.2%]) at median time to LR of 1.1 years (range, 0.11-5.27). Of 95 eligible for delayed surgery after neoadjuvant therapy, 89 (93.7%) achieved R0/R1 margins. Overall LC after RT were as follows: R0, 106 of 109 (97%); R1, 51 of 60 (85%); and R2/unresectable, 2 of 6 (33%). LR predictors include extent of delayed resection (P <.001), imaging response before delayed surgery (P < .001), histologic subtype (P <.001), and no RT (P = .046). The 5-year event-free survival was significantly lower (P = .0003) for patients unable to undergo R0/R1 resection. CONCLUSIONS: Risk-based treatment for young patients with high-grade NRSTS treated on ARST0332 produced very high LC, particularly after R0 resection (97%), despite lower-than-standard RT doses. Neoadjuvant CT/RT enabled delayed R0/R1 resection in most patients and is preferred over adjuvant therapy due to the lower RT dose delivered.

Considerations for the use of imaging tools for phase II treatment trials in oncology.

In the context of assessing tumor response, imaging tools have the potential to play a vital role in phase II and III treatment trials. If the imaging test is able to predict potential phase III success in a reliable fashion, it would be a useful tool in phase II trial design as it may provide for a more rapid and timely response assessment. The benefits and challenges of using anatomic imaging measures as well as the promising molecular imaging measures, primarily fluorodeoxyglucose-positron emission tomography, are discussed here. The general issues related to successful implementation of advanced imaging in the context of phase II treatment trials are discussed.

The rates of second lung cancers and the survival of surgically-resected second primary lung cancers in patients undergoing resection of an initial primary lung cancer.

BACKGROUND: The Lung Cancer Screening Trial demonstrated improved overall survival (OS) and lung cancer specific survival (LCSS), likely due to finding early-stage NSCLC. The purpose of our investigation is to evaluate whether long-term surveillance strategies (4+ years after surgical resection of the initial lung cancer(1LC)) would be beneficial in NSCLC patients by assessing the rates of second lung cancers(2LC) and the OS/LCSS in patients undergoing definitive surgery in 1LC as compared to 2LC (>48 months after 1LC) populations. METHODS: SEER13/18 database was reviewed for patients during 1998-2013. Log-rank tests were used to determine the OS/LCSS differences between the 1LC and 2LC in the entire surgical group(EG) and in those having an early-stage resectable tumors (ESR, tumors <4 cm, node negative). Joinpoint analysis was used to determine rates of second cancers 4-10 year after 1LC using SEER-9 during years 1985-2014. RESULTS: The rate of 2LCs was significantly less than all other second cancers until 2001 when the incidence of 2LCs increased sharply and became significantly greater than all other second cancers in females starting in year 2005 and in men starting in year 2010. OS/LCSS, adjusted for propensity score by using inverse probability weighting, demonstrated similar OS, but worse LCSS for 2LCs in the EG, but similar OS/LCSSs in the ESR group. CONCLUSION: Because the rate of 2LCs are increasing and because the OS/LCSS of the 1LC and 2LC are similar in early-stage lesions, we feel that continued surveillance of patients in order to find early-stage disease may be beneficial.

Developing a multi-institutional PACS archive and designing processes to manage the shift from a film to a digital-based archive.

The Quality Assurance Review Center (QARC) works to improve the standards of care in treating cancer by improving the quality of clinical trials medicine. QARC operates as a data management and review center providing quality assurance services for multiple external groups including cooperative groups and pharmaceutical companies. As the medical world migrates from analog film to digital files, QARC has developed an innovative and unique digital imaging management system to accommodate this trend. As QARC acquires electronic data from institutions across six continents, the system is continually developed to accommodate Digital Imaging and Communications in Medicine (DICOM) imaging originating from a wide variety of Picture Archival and Communications System (PACS) manufacturers, thus creating one of the largest and most diverse multi-institutional imaging archives in the cancer research community.

Processes for quality improvements in radiation oncology clinical trials.

Quality assurance in radiotherapy (RT) has been an integral aspect of cooperative group clinical trials since 1970. In early clinical trials, data acquisition was nonuniform and inconsistent and computational models for radiation dose calculation varied significantly. Process improvements developed for data acquisition, credentialing, and data management have provided the necessary infrastructure for uniform data. With continued improvement in the technology and delivery of RT, evaluation processes for target definition, RT planning, and execution undergo constant review. As we move to multimodality image-based definitions of target volumes for protocols, future clinical trials will require near real-time image analysis and feedback to field investigators. The ability of quality assurance centers to meet these real-time challenges with robust electronic interaction platforms for imaging acquisition, review, archiving, and quantitative review of volumetric RT plans will be the primary challenge for future successful clinical trials.

Interruptions of once-daily thoracic radiotherapy do not correlate with outcomes in limited stage small cell lung cancer: analysis of CALGB phase III trial 9235.

PURPOSE: Retrospective data suggests prolonging the time to complete thoracic radiotherapy (TRT) may negatively impact tumor control and survival in limited stage small cell lung cancer (LSCLC). We examined the association between TRT duration and outcomes on a prospective phase III study. MATERIAL AND METHODS: This review included 267 patients who received protocol TRT on a phase III CALGB LSCLC study assessing the addition of tamoxifen to standard chemo-radiotherapy. TRT, to a planned dose of 50Gy in 2Gy daily fractions, was initiated with the fourth chemotherapy cycle. TRT interruptions were mandated for hematologic toxicity (granulocytes<1000/mm3 or platelets<75,000/mm3) and esophageal toxicity (dysphagia necessitating intravenous hydration). RESULTS: TRT interruptions > or =3 days occurred in 115 patients (43%), most frequently during the 4th week of TRT, and did not differ between treatment arms. Hematologic toxicity and esophageal toxicity were the most frequent indications for interrupting TRT. Variables including advanced age (>70 years), gender, race, or radiotherapy treatment volume did not predict for TRT interruptions. Overall survival (OS) and local tumor control did not correlate with the administration of TRT interruptions or with TRT duration. CONCLUSION: Toxicity mandated interruptions of conventional dose, once-daily, TRT may not adversely affect outcomes for patients receiving TRT concurrent with chemotherapy (cycle 4) for LSCLC. The implications for accelerated or high dose TRT regimens are not clear.

Radiation therapy toxicity to the skin.

Radiation therapy has been integral to cancer patient care. The skin is an intentional and unintentional target of therapy, and is sensitive to the volume of normal tissue in the radiation therapy treatment field, daily treatment dose (fractionation), and total treatment dose. We must understand the relationship of these factors to patient outcome as we move toward hypofractionation treatment strategies (radiosurgery). Chemotherapy agents and prescription medications may influence therapy-associated sequelae. Understanding this may prevent significant injury and discomfort. This article reviews established platforms of radiation therapy and sequelae associated with skin therapy. Interactions with other agents and possible predisposition to sequelae are reviewed. Skin cancer resulting from treatment and disease processes associated with possible limited outcome are also reviewed.

Effect of Standard Radiotherapy With Cisplatin vs Accelerated Radiotherapy With Panitumumab in Locoregionally Advanced Squamous Cell Head and Neck Carcinoma: A Randomized Clinical Trial.

IMPORTANCE: The Canadian Cancer Trials Group study HN.6 is the largest randomized clinical trial to date comparing the concurrent administration of anti-epidermal growth factor receptor (EGFR) monoclonal antibodies with radiotherapy (RT) to standard chemoradiotherapy in locoregionally advanced squamous cell carcinoma of the head and neck (LA-SCCHN). OBJECTIVE: To compare progression-free survival (PFS) in patients with LA-SCCHN treated with standard-fractionation RT plus high-dose cisplatin vs accelerated-fractionation RT plus the anti-EGFR antibody panitumumab. DESIGN, SETTING, AND PARTICIPANTS: A randomized phase 3 clinical trial in 17 Canadian centers. A total of 320 patients were randomized between December 2008 and November 2011. INTERVENTIONS: Patients with TanyN+M0 or T3-4N0M0 LA-SCCHN were randomized 1:1 to receive standard-fractionation RT (70 Gy/35 over 7 weeks) plus cisplatin at 100 mg/m2 intravenous for 3 doses (arm A) vs accelerated-fractionation RT (70 Gy/35 over 6 weeks) plus panitumumab at 9 mg/kg intravenous for 3 doses (arm B). MAIN OUTCOMES AND MEASURES: Primary end point was PFS. Due to an observed declining event rate, the protocol was amended to a time-based analysis. Secondary end points included overall survival, local and regional PFS, distant metastasis-free survival, quality of life, adverse events, and safety. RESULTS: Of 320 patients randomized (268 [84%] male; median age, 56 years), 156 received arm A and 159 arm B. A total of 93 PFS events occurred. By intention-to-treat, 2-year PFS was 73% (95% CI, 65%-79%) in arm A and 76% (95% CI, 68%-82%) in arm B (hazard ratio [HR], 0.95; 95% CI, 0.60-1.50; P = .83). The upper bound of the HR 95% CI exceeded the prespecified noninferiority margin. Two-year overall survival was 85% (95% CI, 78%-90%) in arm A and 88% (95% CI, 82%-92%) in arm B (HR, 0.89; 95% CI, 0.54-1.48; P = .66). Incidence of any grade 3 to 5 nonhematologic adverse event was 88% in arm A and 92% in arm B (P = .25). CONCLUSIONS AND RELEVANCE: With a median follow-up of 46 months, the PFS of panitumumab plus accelerated-fractionation RT was not superior to cisplatin plus standard-fractionation RT in LA-SCCHN and noninferiority was not proven. Despite having negative results, HN.6 has contributed important data regarding disease control and toxic effects of these treatment strategies. TRIAL REGISTRATION: clinicaltrials.gov Identifier: NCT00820248.

Radiotherapy in pediatric medulloblastoma: quality assessment of Pediatric Oncology Group Trial 9031.

PURPOSE: To evaluate the potential influence of radiotherapy quality on survival in high-risk pediatric medulloblastoma patients. METHODS AND MATERIALS: Trial 9031 of the Pediatric Oncology Group (POG) aimed to study the relative benefit of cisplatin and etoposide randomization of high-risk patients with medulloblastoma to preradiotherapy vs. postradiotherapy treatment. Two-hundred and ten patients were treated according to protocol guidelines and were eligible for the present analysis. Treatment volume (whole brain, spine, posterior fossa, and primary tumor bed) and dose prescription deviations were assessed for each patient. An analysis of first site of failure was undertaken. Event-free and overall survival rates were calculated. A log-rank test was used to determine the significance of potential survival differences between patients with and without major deviations in the radiotherapy procedure. RESULTS: Of 160 patients who were fully evaluable for all treatment quality parameters, 91 (57%) had 1 or more major deviations in their treatment schedule. Major deviations by treatment site were brain (26%), spinal (7%), posterior fossa (40%), and primary tumor bed (17%). Major treatment volume or total dose deviations did not significantly influence overall and event-free survival. CONCLUSIONS: Despite major treatment deviations in more than half of fully evaluable patients, underdosage or treatment volume misses were not associated with a worse event-free or overall survival.

The effect of radiation therapy on normal tissue function.

As more patients are treated for their primary malignancy with cure or increased disease-free intervals, injury to normal tissues will become more detectable and an important endpoint for study. Future protocols will probably be modified based on toxicity endpoints. In Hodgkin's disease, current protocols use response-based treatment strategies to limit therapy. The objective is to provide the same level of tumor control and follow normal tissue endpoints for outcome analysis. DVH analysis has improved the ability to analyze endpoint data for normal tissues. These image-guided platforms will provide the infrastructure needed to continue efforts in improving the delivery of radiation therapy.

Induction of malignant transformation of cocultivated hematopoietic stem cells by X-irradiation of murine bone marrow stromal cells in vitro.

X-irradiation of purified primary cultures of mouse bone marrow stroma or permanent cloned marrow stromal cell lines in plateau phase decreases production of macrophage progenitor cell-specific colony-stimulating factor to a plateau minimum of 40% of control levels after doses of 50 to 500 Gy delivered at 2 Gy/min. After 50 Gy there is increased bioavailability of another growth factor(s) that is distinct from macrophage progenitor cell-specific colony-stimulating factor, granulocyte-macrophage progenitor cell colony-stimulating factor, or colony-stimulating factor for multipotential hematopoietic stem cells (interleukin 3). Liquid-phase cocultivation of irradiated stromal cells with either nonadherent cells from continuous marrow cultures or cloned dual granulocyte-macrophage progenitor cell colony-stimulating factor/interleukin 3-dependent hematopoietic progenitor cell lines induces evolution over 5 weeks of factor-independent colony-forming cells. Subcultured factor-independent colonies generated clonal malignant cell lines with multiple distinct karyotypic alterations. Inoculation of 10(6) cells s.c. from factor-independent clones into syngeneic mice produces local granulocytic monomyeloid tumors with spread to spleen, lymph nodes, and bone marrow. These data provide the first demonstration in vitro of indirect X-irradiation leukemogenesis through cells of the marrow stroma.

Induction chemotherapy with docetaxel, cisplatin, fluorouracil, and leucovorin for squamous cell carcinoma of the head and neck: a phase I/II trial.

PURPOSE: A phase I/II trial of docetaxel, cisplatin, fluorouracil (5-FU), and leucovorin (TPFL5) induction chemotherapy for patients with locally advanced squamous cell carcinoma of the head and neck (SCCHN). PATIENTS AND METHODS: Twenty-three previously untreated patients with stage III or IV SCCHN and Eastern Cooperative Oncology Group functional status less than or equal to 2 were treated with TPFL5. Postchemotherapy home support included intravenous fluids, prophylactic antibiotics, and granulocyte colony-stimulating factor (G-CSF). Docetaxel dose was escalated to determine the maximum-tolerated dose (MTD). Fifteen patients were treated with three cycles of TPFL5 at MTD. Patients who achieved either a partial response (PR) or complete response (CR) to three cycles of TPFL5 then received definitive twice-daily radiation therapy. Toxicity and clinical and pathologic response to TPFL5 were assessed. RESULTS: Twenty-three patients received a total of 69 cycles of TPFL5. The MTD was determined to be docetaxel 60 mg/m2. Dose-limiting toxicity (DLT) was neutropenia. Additional significant toxicities at MTD were nausea, mucositis, diarrhea, peripheral neuropathy, and sodium-wasting nephropathy. The overall response rate to TPFL5 was 100%, which included 14 of 23 (61%) clinical CRs and nine of 23 (39%) clinical PRs. Primary-site clinical and pathologic CR rates were 19 of 22 (86%) CRs and 20 of 22 (91%) CRs, respectively. Eight patients had less than a CR in the neck to chemotherapy and, therefore, had postradiation neck dissections, four of which were positive for residual tumor. CONCLUSION: TPFL5 is a tolerable induction regimen in patients with good performance status. The DLT is neutropenia with significant mucositis, diarrhea, peripheral neuropathy, and sodium-wasting nephropathy. The high response rates to TPFL5 justify further evaluation of this combination of agents in the context of formal clinical trials.

Analysis of axillary coverage during tangential radiation therapy to the breast.

PURPOSE: To evaluate the percent of the prescribed radiation dose to the breast delivered to the axillary tissue and to evaluate the volume of the axilla receiving 95% of the prescribed dose with normal and with high tangential fields. METHODS AND MATERIALS: Computed tomographic scan images with 5-mm sections were retrospectively analyzed for 35 patients who had undergone three-dimensional (3D) planning for whole-breast radiation. The axillary nodal region was identified and divided into Levels I to III and Rotter's nodes (RN). Digitally reconstructed radiographs were created, and two plans were developed: (a) the standard clinical opposed tangential irradiation fields and (b) the high-tangential irradiation fields. Axillary coverage was examined by use of dose-volume histograms (DVH), and the average coverage for the four nodal groups was obtained. RESULTS: The data show that with the standard tangential irradiation fields, the average dose delivered to Levels I, II, III, and RN is 66% (standard deviation, or SD = 13%), 44% (SD = 18%), 31% (SD = 20%), and 70% (SD = 19%) of the prescribed dose, respectively. The coverage increases to 86% (SD = 9%), 71% (SD = 19%), 73% (SD = 17%), and 94% (SD = 8%) of the prescribed dose, respectively, for Levels I, II, III, and RN when the high tangential irradiation fields are used. 51% of Level I, 26% of Level II, and 15% of Level III receive 95% of the prescribed dose with normal tangents. The volume increases to 79%, 51%, and 49% of Levels I, II, and III, respectively, with high tangents. CONCLUSION: The tangential fields designed to treat only the breast do not adequately cover the axillary region and, therefore, cannot be relied upon for prophylactic therapy of the axilla. The high tangential irradiation fields increase the dosages received by the axillary region, but the average dosages received by the lower axillary regions are still less than 90% of the prescribed dose.

MKW, a novel hematopoietic antigen.

A novel hematopoietic antigen was identified using a murine monoclonal antibody raised against KG-1 cells. This antigen, termed MKW, was also detected on the surface of the monocytic cell line U937, but not on the K562, ML1, or HL-60 cell lines. On normal hematopoietic cells, the antigen is expressed on the surface of monocytic and myelocytic cells and on a subpopulation of B-cells. During normal hematopoiesis, the surface expression of MKW is greatest and occurs very early on monocytic cells. Alternatively, in myeloid cells, surface expression occurs later and cell maturation is correlated with increased surface expression. When U937 cells are induced to differentiate, surface expression is transiently up-regulated. Surface expression of MKW, however, does not appear to be an activation antigen since activation of purified T- or B-cells failed to increase MKW on the cell surface. Leukemic blasts from 22 of 80 children (27%) with acute myeloblastic leukemia and from 29 of 225 children (13%) with acute lymphoblastic leukemia expressed MKW on the cell surface. Although surface expression of MKW was absent on T-cell lines, peripheral T-cells, and most B-cells, the antigen was identified in the cytoplasm of some B-cells, T-cells, and cell lines. Immunoprecipitation studies showed that MKW is a 52-kDa protein whether expressed on the cell surface or in the cytoplasm, and it appears to be nonglycosylated. Furthermore, studies with phosphatidylinositol-phospholipase C suggested that MKW is not attached to a glycolipid anchor. The biochemical characterization of MKW and its pattern of expression are distinct from any of the previously identified CD groups or published antigens. Since this unique antigen has prognostic significance in leukemia and appears to be associated with cell differentiation, its exact role in hematopoiesis should be investigated.

Macrophage and lymphocyte functions are down-regulated in newborn rabbits.

We have developed an animal model for congenital syphilis. Treponema pallidum is injected intravenously into pregnant rabbits and fetuses are infected in utero. As a prelude to characterizing the immunologic consequences of fetal infection, it was necessary to expand on the baseline information about newborn rabbit immune capabilities. Studies were undertaken to determine splenic macrophage and T lymphocyte functions with emphasis on newer immunologic parameters. Newborns aged 2 weeks were compared to adults. Macrophage capabilities in newborn rabbits differed from those of their adult counterparts. These cells produced similar basal levels of interleukin 1 (IL-1) but failed to respond to the IL-1 stimulants of lipopolysaccharide (LPS) or T. pallidum. Macrophages also exhibited diminished levels of la expression and increased levels of prostaglandin E2 (PGE2) secretion. T lymphocyte functions were altered in newborn spleen preparations. Following concanavalin A (Con A) stimulation, interferon gamma production was half that of adults; in direct contrast, IL-2 production was twice that of adults. Con A-induced lymphocyte proliferation was markedly decreased in newborn preparations. This diminished response resulted from down-regulation rather than immaturity. When newborn splenic cells were stimulated with Con A in the presence of indomethacin, anti-transforming growth factor (anti-TGF), or exogenous IL-1/IL-2, better proliferation resulted. PGE2, which is well established as a down-regulator of newborn immune functions in human and mouse systems, also appears to play a role in suppressing newborn rabbit functions. TGF is a potent suppressor of a number of adult immunologic reactions. This is the first documentation of the potential role of this factor in down-regulating newborn immune capabilities. These findings provide a framework for future investigations of our congenital syphilis model.