Clinical validation of deep learning algorithms for radiotherapy targeting of non-small-cell lung cancer: an observational study.

BACKGROUND: Artificial intelligence (AI) and deep learning have shown great potential in streamlining clinical tasks. However, most studies remain confined to in silico validation in small internal cohorts, without external validation or data on real-world clinical utility. We developed a strategy for the clinical validation of deep learning models for segmenting primary non-small-cell lung cancer (NSCLC) tumours and involved lymph nodes in CT images, which is a time-intensive step in radiation treatment planning, with large variability among experts. METHODS: In this observational study, CT images and segmentations were collected from eight internal and external sources from the USA, the Netherlands, Canada, and China, with patients from the Maastro and Harvard-RT1 datasets used for model discovery (segmented by a single expert). Validation consisted of interobserver and intraobserver benchmarking, primary validation, functional validation, and end-user testing on the following datasets: multi-delineation, Harvard-RT1, Harvard-RT2, RTOG-0617, NSCLC-radiogenomics, Lung-PET-CT-Dx, RIDER, and thorax phantom. Primary validation consisted of stepwise testing on increasingly external datasets using measures of overlap including volumetric dice (VD) and surface dice (SD). Functional validation explored dosimetric effect, model failure modes, test-retest stability, and accuracy. End-user testing with eight experts assessed automated segmentations in a simulated clinical setting. FINDINGS: We included 2208 patients imaged between 2001 and 2015, with 787 patients used for model discovery and 1421 for model validation, including 28 patients for end-user testing. Models showed an improvement over the interobserver benchmark (multi-delineation dataset; VD 0·91 [IQR 0·83-0·92], p=0·0062; SD 0·86 [0·71-0·91], p=0·0005), and were within the intraobserver benchmark. For primary validation, AI performance on internal Harvard-RT1 data (segmented by the same expert who segmented the discovery data) was VD 0·83 (IQR 0·76-0·88) and SD 0·79 (0·68-0·88), within the interobserver benchmark. Performance on internal Harvard-RT2 data segmented by other experts was VD 0·70 (0·56-0·80) and SD 0·50 (0·34-0·71). Performance on RTOG-0617 clinical trial data was VD 0·71 (0·60-0·81) and SD 0·47 (0·35-0·59), with similar results on diagnostic radiology datasets NSCLC-radiogenomics and Lung-PET-CT-Dx. Despite these geometric overlap results, models yielded target volumes with equivalent radiation dose coverage to those of experts. We also found non-significant differences between de novo expert and AI-assisted segmentations. AI assistance led to a 65% reduction in segmentation time (5·4 min; p<0·0001) and a 32% reduction in interobserver variability (SD; p=0·013). INTERPRETATION: We present a clinical validation strategy for AI models. We found that in silico geometric segmentation metrics might not correlate with clinical utility of the models. Experts' segmentation style and preference might affect model performance. FUNDING: US National Institutes of Health and EU European Research Council.

A Phase 1 Dose-Escalation Trial of Radiation Therapy and Concurrent Cisplatin for Stage II and III Triple-Negative Breast Cancer.

PURPOSE: Patients with triple-negative breast cancer (TNBC) experience higher local-regional recurrence rates than those with luminal or HER2-positive tumors. This prospective, phase 1B trial was designed to assess the safety and to establish the maximum tolerated dose (MTD) of cisplatin with radiation therapy for women with early-stage TNBC. METHODS AND MATERIALS: Eligible patients had stage II or III TNBC. Cisplatin was initiated at 10 mg/m2 intravenously once weekly during radiation and then escalated in a 3 + 3 design by 10 mg/m2 at each dose level until 40 mg/m2, or the MTD, was reached. Patients undergoing breast-conserving therapy (BCT) or mastectomy were accrued in separate parallel cohorts during dose escalation, followed by a 10-patient expansion at the MTD. RESULTS: During 2013 to 2018, 55 patients were accrued. Four patients developed dose-limiting toxicity. In the BCT cohort, 1 patient receiving 40 mg/m2 developed tinnitus resulting in a cisplatin delay; therefore, this was the BCT cohort MTD. In the mastectomy cohort, 1 patient receiving 20 mg/m2 developed a grade 3 urinary infection, and 2 additional patients had dose-limiting toxicities at 40 mg/m2 (grade 3 neutropenia and grade 2 tinnitus), both resulting in cisplatin delay. Thus, 30 mg/m2 was the mastectomy cohort MTD. Median follow-up was 48.5 months. Three-year disease-free survival was 74.7% for the BCT cohort and 64.4% for the mastectomy cohort. CONCLUSIONS: Adjuvant radiation therapy with concurrent cisplatin is feasible with a recommended phase 2 dose of 30 mg/m2 and 40 mg/m2 intravenously weekly in mastectomy and BCT cohorts, respectively.

Feasibility of hippocampal avoidance whole brain radiation in patients with hippocampal involvement: Data from a prospective study.

PURPOSE: Among patients with brain metastases, hippocampal avoidance whole brain radiation (HA-WBRT) preserves neurocognitive function relative to conventional WBRT but the feasibility of hippocampal sparing in patients with metastases in/near the hippocampus is unknown. We identified the incidence of hippocampal/perihippocampal metastases and evaluated the feasibility of HA-WBRT in such patients. MATERIALS/METHODS: Dosimetric data from 34 patients randomized to HA-WBRT (30 Gy/10 fractions) in a phase III trial (NCT03075072) comparing HA-WBRT to stereotactic radiation in patients with 5 to 20 brain metastases were analyzed. Patients with metastases in/near the hippocampi received HA-WBRT with prioritization of tumor coverage over hippocampal avoidance. Target coverage and hippocampal sparing metrics were compared between patients with targets in/near the hippocampus versus not. RESULTS: In total, 9 of 34 (26%) patients had targets in the hippocampus and an additional 5 of 34 (15%) patients had targets in the hippocampal avoidance zone (HAZ, hippocampus plus 5 mm expansion) but outside the hippocampus. Patients with targets within the hippocampus and those with targets in the HAZ but outside the hippocampus were spared 34% and 73% of the ipsilateral mean biologically equivalent prescription dose, respectively. Of the latter cohort, 88% and 25% met conventional hippocampal sparing metrics of Dmin ≤ 9 Gy and Dmax ≤ 16 Gy, respectively. Among 11 patients with unilateral hippocampal/perihippocampal involvement, the uninvolved/contralateral hippocampus was limited to Dmin ≤ 9 Gy and Dmax ≤ 17 Gy in all cases. CONCLUSIONS: In this study, a substantial percentage of patients with 5 to 20 brain metastases harbored metastases in/near the hippocampus. In such cases, minimizing hippocampal dose while providing tumor coverage was feasible and may translate to neurocognitive protection.

Local control after brain-directed radiation in patients with cystic versus solid brain metastases.

PURPOSE: Brain metastases can be radiographically cystic or solid. Cystic metastases are associated with a greater intracranial disease burden and poorer oncologic outcomes, but the impact of cystic versus solid appearance on local control after radiation remains unknown. We investigated whether cystic versus solid nature is predictive of local control after management with stereotactic or whole brain radiation (WBRT) and whether the radiation modality utilized is an effect modifier. METHODS: We identified 859 patients with 2211 newly-diagnosed brain metastases managed with upfront stereotactic radiation or WBRT without preceding resection/aspiration at Brigham and Women's Hospital/Dana-Farber Cancer Institute between 2000 and 2015. Multivariable Cox regression with an interaction term and sandwich covariance matrix was used to quantify local failure. RESULTS: Cystic lesions were more likely to recur than solid ones when managed with stereotactic radiation (HR 2.33, 95% CI 1.32-4.10, p = 0.004) but not WBRT (HR 0.92, 95% CI 0.62-1.36, p = 0.67), p-interaction = 0.007. 1 year local control rates for cystic versus solid metastases treated with stereotactic radiation were 75% versus 88%, respectively; estimates with WBRT were 76% versus 76%, respectively. However, no significant differences were noted between the two cohorts in post-radiation outcomes including all-cause mortality and neurologic death (p > 0.05). CONCLUSIONS: Among patients with brain metastases, stereotactic radiation yields improved local control and less morbidity than WBRT, and consequently for many patients the cystic versus solid designation does not impact treatment selection. However, our results suggest that in patients with a large number of cystic brain metastases, a lower threshold to consider WBRT, as opposed to stereotactic radiation, should be employed. If our results can be confirmed, further investigation into the underlying mechanism(s) would be warranted.

Outcomes following radiation for cutaneous squamous cell carcinoma of the head and neck: Associations between immune suppression and recurrence.

BACKGROUND: Data evaluating outcomes and patterns of recurrence following radiation therapy (RT) for cutaneous squamous cell carcinoma (cSCC) of the head and neck are limited. METHODS: We performed a retrospective analysis of 111 head and neck cSCC patients treated with RT at 4 affiliated institutions. RESULTS: With median follow-up of 7 months, there were 29 (26%) recurrences, 73% of which were nodal (n = 21). Immunosuppression (IS) was the only factor associated with recurrence (47% in IS, 22% in non-IS, P = .04), and also with time to recurrence in multivariate analysis (HR 5.5; P = .03). No factors were associated with recurrence among patients who received definitive RT. The majority of patients who recurred were salvaged with surgery (n = 20, 69%). CONCLUSION: In a cohort of cSCC treated with radiotherapy, there was an association between IS and increased failure risk. The majority of failures were salvaged surgically.

Neurosurgical Resection and Stereotactic Radiation Versus Stereotactic Radiation Alone in Patients with a Single or Solitary Brain Metastasis.

BACKGROUND: Brain metastases commonly manifest in patients with cancer, with ∼20%-50% presenting with 1 intracranial lesion. Among patients with 1, small brain metastasis and controlled or absent extracranial disease, it remains unclear whether aggressive intracranial management using neurosurgical resection plus cavity stereotactic radiosurgery/stereotactic radiotherapy (SRS/SRT) rather than SRS/SRT alone is beneficial. In patients with controlled or absent extracranial disease and 1 brain metastasis ≤2 cm in size, we evaluated the effect of surgery plus SRS/SRT compared with SRS/SRT on oncologic outcomes, including overall survival. METHODS: We retrospectively identified 86 patients with controlled or absent extracranial disease and 1 brain metastasis ≤2 cm in size who had been treated from 2000 to 2015 at our institution. We examined differences in the rates of local and distant failure, use of salvage treatment, and other oncologic outcomes, including all-cause mortality. RESULTS: The baseline characteristics were similar between the 2 cohorts. The median follow-up period for the surviving patients was 38 months. On multivariable analysis, surgical resection plus cavity SRS/SRT was associated with a lower risk of all-cause mortality (hazard ratio, 0.44; 95% confidence interval, 0.19-1.00; P = 0.05) compared with SRS/SRT alone. The 1- and 2-year rates of overall survival were 100% and 88% versus 74% and 52% for surgery plus cavity SRS/SRT versus SRS/SRT alone, respectively. CONCLUSIONS: Aggressive, local therapy, including neurosurgical resection, might benefit patients with 1 brain metastasis in the context of controlled or absent systemic disease, even if the lesion in question is small. Further studies are needed to evaluate these associations.

Proton Stereotactic Radiosurgery for Brain Metastases: A Single-Institution Analysis of 370 Patients.

PURPOSE: To report the first series of proton stereotactic radiosurgery (SRS) for the treatment of patients with single or multiple brain metastases, including failure patterns, survival outcomes, and toxicity analysis. METHODS AND MATERIALS: This was a single-institution, retrospective study of 815 metastases from 370 patients treated with proton SRS between April 1991 and November 2016. Cumulative incidence estimates of local failure, distant brain failure, and pathologically confirmed radionecrosis and Kaplan-Meier estimates of overall survival were calculated. Fine and Gray and Cox regressions were performed to ascertain whether clinical and treatment factors were associated with the described endpoints. RESULTS: The median follow-up from proton SRS was 9.2 months. The 6- and 12-month estimates of local failure, distant brain failure, and overall survival were 4.3% (95% confidence interval [CI] 3.0%-5.9%) and 8.5% (95% CI 6.7%-10.6%), 39.1% (95% CI 34.1%-44.0%) and 48.2% (95% CI 43.0%-53.2%), and 76.0% (95% CI 71.3%-80.0%) and 51.5% (95% CI 46.3%-56.5%), respectively. The median survival was 12.4 months (95% CI 10.8-14.0 months) after proton SRS. The most common symptoms were low-grade fatigue (12.5%), headache (10.0%), motor weakness (6.2%), seizure (5.8%), and dizziness (5.4%). The rate of pathologically confirmed radionecrosis at 12 months was 3.6% (95% CI 2.0%-5.8%), and only target volume was associated on multivariate analysis (subdistribution hazard ratio 1.13, 95% CI 1.0-1.20). CONCLUSIONS: To the best of our knowledge, this is the first reported series of proton SRS for the management of brain metastases. Moderate-dose proton SRS is well tolerated and can achieve good local control outcomes, comparable to those obtained with conventional photon SRS strategies. Although proton SRS remains resource-intensive, future strategies evaluating its selective utility in patients who would benefit most from integral dose reduction should be explored.

Radiation therapy in the management of pituitary adenomas.

Radiation therapy in the form of fractionated treatment or radiosurgery has an important role in the management of pituitary adenomas. Radiation is a reliable way of gaining local control for radiographically progressing pituitary adenomas. For functioning adenomas that are biochemically recurrent or persistent, radiation therapy is less consistent in offering biochemical normalization and often requires a latency period of years or decades. The decision of when to use radiation therapy is a delicate balance between its benefits and late sequelae, especially in the context of benign disease. Recent technological advances in radiation oncology hold the potential to minimize dose to uninvolved normal tissue and therefore reduce the risk of toxicity.

The impact of PSA and digital rectal examination on the risk of prostate cancer specific mortality in men with a PSA level <2.5 ng/ml.

INTRODUCTION: It is unknown whether a normal range, diagnostic serum prostate specific antigen (PSA) level's influence on prostate cancer specific mortality (PCSM) is dependent upon digital rectal examination (DRE) findings. METHODS: Between 2004 and 2007, 9081 men diagnosed with non-palpable (T1c, N=1710) or palpable (T2-T4, N=7371) and non-metastatic prostate cancer (PC) were identified from surveillance, epidemiology, and end results data, selected based on pre-treatment PSA<2.5 ng/ml. A multivariable competing risks regression model evaluated whether DRE findings interacted with PSA level in predicting risk of PCSM. RESULTS: After median follow-up of 2.83 years, 118 of 548 deaths (21.5%) were due to PC. Increasing diagnostic PSA was associated with increased risk of PCSM (AHR=3.52; 95% CI: 1.25-9.89; P=.017) in men with T1c, Gleason score 7-10 PC, but decreased PCSM risk (AHR=0.66; 95% CI: 0.52-0.83; P<.001) for men with T2-T4 PC and any Gleason score. DISCUSSION: For men with diagnostic PSA level <2.5 ng/ml and palpable PC, risk of early PCSM increases by 34% for a 1 point decrease in PSA from 2. This suggests the existence of clinically detectable, low PSA secreting disease with an elevated risk of early PCSM, highlighting the importance of the DRE in men with PC and normal range, diagnostic PSA.

Long-term cosmetic outcomes and toxicities of proton beam therapy compared with photon-based 3-dimensional conformal accelerated partial-breast irradiation: a phase 1 trial.

PURPOSE: To present long-term outcomes of a prospective feasibility trial using either protons or 3-dimensional conformal photon-based (accelerated partial-breast irradiation [APBI]) techniques. METHODS AND MATERIALS: From October 2003 to April 2006, 98 evaluable patients with stage I breast cancer were treated with APBI (32 Gy in 8 fractions given twice daily) on a prospective clinical trial: 19 with proton beam therapy (PBT) and 79 with photons or mixed photons/electrons. Median follow-up was 82.5 months (range, 2-104 months). Toxicity and patient satisfaction evaluations were performed at each visit. RESULTS: At 7 years, the physician rating of overall cosmesis was good or excellent for 62% of PBT patients, compared with 94% for photon patients (P=.03). Skin toxicities were more common for the PBT group: telangiectasia, 69% and 16% (P=.0013); pigmentation changes, 54% and 22% (P=.02); and other late skin toxicities, 62% and 18% (P=.029) for PBT and photons, respectively. There were no significant differences between the groups in the incidences of breast pain, edema, fibrosis, fat necrosis, skin desquamation, and rib pain or fracture. Patient-reported cosmetic outcomes at 7 years were good or excellent for 92% and 96% of PBT and photon patients, respectively (P=.95). Overall patient satisfaction was 93% for the entire cohort. The 7-year local failure rate for all patients was 6%, with 3 local recurrences in the PBT group (7-year rate, 11%) and 2 in photon-treated patients (4%) (P=.22). CONCLUSIONS: Local failure rates of 3-dimensional APBI and PBT were similar in this study. However, PBT, as delivered in this study, led to higher rates of long-term telangiectasia, skin color changes, and skin toxicities. We recommend the use of multiple fields and treatment of all fields per treatment session or the use of scanning techniques to minimize skin toxicity.

DNA repair pathway gene expression score correlates with repair proficiency and tumor sensitivity to chemotherapy.

Mutagenesis is a hallmark of malignancy, and many oncologic treatments function by generating additional DNA damage. Therefore, DNA damage repair is centrally important in both carcinogenesis and cancer treatment. Homologous recombination (HR) and nonhomologous end joining are alternative pathways of double-strand DNA break repair. We developed a method to quantify the efficiency of DNA repair pathways in the context of cancer therapy. The recombination proficiency score (RPS) is based on the expression levels for four genes involved in DNA repair pathway preference (Rif1, PARI, RAD51, and Ku80), such that high expression of these genes yields a low RPS. Carcinoma cells with low RPS exhibit HR suppression and frequent DNA copy number alterations, which are characteristic of error-prone repair processes that arise in HR-deficient backgrounds. The RPS system was clinically validated in patients with breast or non-small cell lung carcinomas (NSCLCs). Tumors with low RPS were associated with greater mutagenesis, adverse clinical features, and inferior patient survival rates, suggesting that HR suppression contributes to the genomic instability that fuels malignant progression. This adverse prognosis associated with low RPS was diminished if NSCLC patients received adjuvant chemotherapy, suggesting that HR suppression and associated sensitivity to platinum-based drugs counteract the adverse prognosis associated with low RPS. Therefore, RPS may help oncologists select which therapies will be effective for individual patients, thereby enabling more personalized care.

Acute gastrointestinal toxicity and tumor response with preoperative intensity modulated radiation therapy for rectal cancer.

BACKGROUND: Preoperative chemoradiotherapy (preopCRT) for locally advanced rectal cancer is associated with grade 3 or higher acute gastrointestinal (GI) toxicity. This study was conducted to determine whether intensity-modulated radiation therapy (IMRT) significantly reduces acute GI toxicity, compared to 3-dimensional conformal RT (3D-CRT) in preopCRT for rectal cancer. METHODS: A retrospective analysis was conducted of 48 patients treated between January 2002 and August 2010 with preopCRT for rectal cancer. 3D-CRT or IMRT was administered at a planned dose of 45-50.4 Gy to patients positioned prone on a bowel-displacement device. Data regarding patient and tumor characteristics, treatment, acute toxicity, and tumor response were collected. Comparisons of acute toxicity and treatment response between 3D-CRT and IMRT were performed with the Chi-square or Fisher's exact test. RESULTS: There were no significant differences in radiation dose, median age, race, gender, stage, type of concurrent chemotherapy, pathologic complete response (pCR), or type of surgery (lower anterior or abdominal perineal resection) between 3D-CRT and IMRT. There was a significant reduction in grade 2 or higher GI toxicity (3D-CRT, 60.7%; IMRT, 30%; P = .036) and grade 2 or higher diarrhea (3D-CRT, 42.8%; IMRT, 10%; P = .014). Two patients who underwent 3D-CRT required a treatment break (grade 3 diarrhea and grade 3 dehydration). Radiation duration was significantly less (IMRT, 35 days; 3D-CRT, 39 days; P ≤ .0001). pCR rates were 16.7% for 3D-CRT and 21.4% for IMRT (nonsignificant [NS]); pCR+microscopic residual rates were 57.1% for IMRT and 27.8% for 3D-CRT (P = .093). CONCLUSION: Maximal bowel displacement with IMRT yields favorable acute GI toxicity and pathologic downstaging profiles, as compared to 3D-CRT in preoperative CRT for rectal cancer and warrants further prospective investigation.

Integrated genomic characterization of endometrial carcinoma.

We performed an integrated genomic, transcriptomic and proteomic characterization of 373 endometrial carcinomas using array- and sequencing-based technologies. Uterine serous tumours and ∼25% of high-grade endometrioid tumours had extensive copy number alterations, few DNA methylation changes, low oestrogen receptor/progesterone receptor levels, and frequent TP53 mutations. Most endometrioid tumours had few copy number alterations or TP53 mutations, but frequent mutations in PTEN, CTNNB1, PIK3CA, ARID1A and KRAS and novel mutations in the SWI/SNF chromatin remodelling complex gene ARID5B. A subset of endometrioid tumours that we identified had a markedly increased transversion mutation frequency and newly identified hotspot mutations in POLE. Our results classified endometrial cancers into four categories: POLE ultramutated, microsatellite instability hypermutated, copy-number low, and copy-number high. Uterine serous carcinomas share genomic features with ovarian serous and basal-like breast carcinomas. We demonstrated that the genomic features of endometrial carcinomas permit a reclassification that may affect post-surgical adjuvant treatment for women with aggressive tumours.

Radiotherapy in the management of localized primary cutaneous B-cell lymphoma.

The optimal therapy and radiation dose for patients with localized primary cutaneous B-cell lymphoma (PCBCL) are unknown. We retrospectively identified 23 patients with localized (T1-T2) PCBCL treated with definitive radiation to doses ranging from 30 to 44 Gy (median, 36 Gy). With a median follow-up of 4.8 years, the 5-year overall survival rate was 100%, the relapse-free survival rate was 71% (95% confidence interval, 46-86%) and there were no local recurrences, suggesting that radiotherapy to a dose of 30 Gy may be sufficient for cure.

Primary hyperparathyroidism in adolescents: the same but different.

PURPOSE: Primary hyperparathyroidism has been studied more extensively in adults than in adolescents. The objective of this study is to define the similarities and differences that exist between these groups. METHODS: A retrospective review of 1,000 primary hyperparathyroidism patients undergoing parathyroidectomy at a single tertiary-care university teaching hospital between 1990 and 2004. All patients 20 years of age or younger comprised our study cohort, and were compared to two historical adult groups. RESULTS: Of 1,000 parathyroidectomies, 21 (2.1 %) were 20 years of age or younger (adolescent). The adolescents presented with higher serum calcium levels (p < 0.01) more severe symptoms (p = 0.02), more renal stones (p = 0.048), and a higher incidence of hypercalcemic crisis (p = 0.02), when compared with adults. We found that 67 % suffered from a triad of tiredness, weakness, and depression versus 39 % of adults (p = 0.02). Sestamibi scans were less helpful in the adolescents than in adults. Similar to the adults, 86 % of adolescent patients had single gland disease, and 95 % were cured at the first operation. CONCLUSION: Adolescents with primary hyperparathyroidism typically have more severe disease than adults. Contrary to popular belief, most adolescents have single gland disease and not hyperplasia associated with a genetic disorder.

External beam accelerated partial-breast irradiation using 32 gy in 8 twice-daily fractions: 5-year results of a prospective study.

PURPOSE: External beam accelerated partial breast irradiation (APBI) is an increasingly popular technique for treatment of patients with early stage breast cancer following breast-conserving surgery. Here we present 5-year results of a prospective trial. METHODS AND MATERIALS: From October 2003 through November 2005, 98 evaluable patients with stage I breast cancer were enrolled in the first dose step (32 Gy delivered in 8 twice-daily fractions) of a prospective, multi-institutional, dose escalation clinical trial of 3-dimensional conformal external beam APBI (3D-APBI). Median age was 61 years; median tumor size was 0.8 cm; 89% of tumors were estrogen receptor positive; 10% had a triple-negative phenotype; and 1% had a HER-2-positive subtype. Median follow-up was 71 months (range, 2-88 months; interquartile range, 64-75 months). RESULTS: Five patients developed ipsilateral breast tumor recurrence (IBTR), for a 5-year actuarial IBTR rate of 5% (95% confidence interval [CI], 1%-10%). Three of these cases occurred in patients with triple-negative disease and 2 in non-triple-negative patients, for 5-year actuarial IBTR rates of 33% (95% CI, 0%-57%) and 2% (95% CI, 0%-6%; P<.0001), respectively. On multivariable analysis, triple-negative phenotype was the only predictor of IBTR, with borderline statistical significance after adjusting for tumor grade (P=.0537). CONCLUSIONS: Overall outcomes were excellent, particularly for patients with estrogen receptor-positive disease. Patients in this study with triple-negative breast cancer had a significantly higher IBTR rate than patients with other receptor phenotypes when treated with 3D-APBI. Larger, prospective 3D-APBI clinical trials should continue to evaluate the effect of hormone receptor phenotype on IBTR rates.

Cancer cells harboring MET gene amplification activate alternative signaling pathways to escape MET inhibition but remain sensitive to Hsp90 inhibitors.

Hepatocyte growth factor/scatter factor (HGF/SF) receptor c-Met is implicated in growth, invasion and metastasis of many tumors. Tumor cells harboring MET gene amplification are initially sensitive to c-Met tyrosine kinase inhibitors (TKI), but escape from long-term treatment has not been investigated. C-Met is a client of heat shock protein 90 (Hsp90) and is destabilized by Hsp90 inhibitors, suggesting that these drugs may inhibit tumors driven by MET amplification, although tumor escape under these conditions also has not been explored. Here, we evaluated the initial inhibitory effects of, and the likelihood of escape from, the Hsp90 inhibitor 17-allylamino-17-demethoxygeldanamycin (17-AAG) and the c-Met TKI SU11274, using two cell lines harboring MET gene amplification. 17-AAG inhibited cell growth in both cell lines and induced substantial apoptosis, whereas SU11274 was only growth inhibitory in one cell line. In both cell lines, c-Met-dependent Akt, Erk and/or STAT3 signaling, as well as activation of the EGFR family, resumed shortly after treatment with c-Met TKI despite sustained c-Met inhibition. PKC delta upregulation may participate in reactivation of c-Met downstream signaling in both cell lines. In contrast to c-Met TKI, 17-AAG destabilized c-Met protein and durably blocked reactivation of downstream signaling pathways and EGFR family members. Our data demonstrate that downstream signaling in tumor cells overexpressing c-Met is not stably suppressed by c-Met TKI, even though c-Met remains fully inhibited. In contrast, Hsp90 inhibitors provide long-lasting suppression of c-Met-dependent signaling, and these drugs should be further evaluated in tumors driven by MET gene amplification.

Targeting Hsp90 prevents escape of breast cancer cells from tyrosine kinase inhibition.

Recent studies have identified development of resistance to tyrosine kinase inhibition (TKI) as a significant roadblock to effective treatment. One mechanism of resistance recently appreciated involves 'oncogene switching', or the re-activation of signaling pathways by one or more redundant upstream activators. In breast cancer models, ErbB TKIs such as gefitinib have been shown to lose the ability to modulate ErbB-driven signaling pathways over time, even though ErbB inhibition is maintained. Although incomplete ErbB inhibition has been proposed to underlie this phenomenon, our findings suggest that oncogene switching can also re-activate downstream signaling pathways in breast cancer cells, even when ErbB inhibition is complete. We find that ErbB TKI-induced Src activation mediates downstream signaling rebound in SKBR3 cells, and we show that combination of Src and ErbB inhibitors is more effective and longlasting than is either TKI alone. Finally, the Hsp90 inhibitor 17-AAG, by simultaneously and durably inhibiting multiple signaling activators including ErbB and Src kinases, does not permit oncogene switching and results in a more prolonged and robust inhibition of downstream signaling pathways in breast cancer cells than do individual TKIs. These data support the continued clinical evaluation of Hsp90 inhibitors in breast cancer.