Thermal Bridges on Building Rooftops

Thermal Bridges on Building Rooftops (TBBR) is a multi-channel remote sensing dataset. It was recorded during six separate UAV fly-overs of the city center of Karlsruhe, Germany, and comprises a total of 926 high-resolution images with 6927 manually-provided thermal bridge annotations. Each image provides five channels: three color, one thermographic, and one computationally derived height map channel. The data is pre-split into training and test data subsets suitable for object detection and instance segmentation tasks. All data is organized and structured to comply with FAIR principles, i.e. being findable, accessible, interoperable, and reusable. It is publicly available and can be downloaded from the Zenodo data repository. This work provides a comprehensive data descriptor for the TBBR dataset to facilitate broad community uptake.

Learning Tree Structures from Leaves For Particle Decay Reconstruction

In this work, we present a neural approach to reconstructing rooted tree graphs describing hierarchical interactions, using a novel representation we term the Lowest Common Ancestor Generations (LCAG) matrix. This compact formulation is equivalent to the adjacency matrix, but enables learning a tree’s structure from its leaves alone without the prior assumptions required if using the adjacency matrix directly. Employing the LCAG therefore enables the first end-to-end trainable solution which learns the hierarchical structure of varying tree sizes directly, using only the terminal tree leaves to do so. In the case of high-energy particle physics, a particle decay forms a hierarchical tree structure of which only the final products can be observed experimentally, and the large combinatorial space of possible trees makes an analytic solution intractable. We demonstrate the use of the LCAG as a target in the task of predicting simulated particle physics decay structures using both a Transformer encoder and a Neural Relational Inference encoder Graph Neural Network. With this approach, we are able to correctly predict the LCAG purely from leaf features for a maximum tree-depth of 8 in 92.5% of cases for trees up to 6 leaves (including) and 59.7% for trees up to 10 in our simulated dataset.

Search for a Light Higgs Boson in Single-Photon Decays of ϒ(1S) Using ϒ(2S)→π^{+}π^{-}ϒ(1S) Tagging Method

We search for a light Higgs boson (A^{0}) decaying into a τ^{+}τ^{-} or μ^{+}μ^{-} pair in the radiative decays of ϒ(1S). The production of ϒ(1S) mesons is tagged by ϒ(2S)→π^{+}π^{-}ϒ(1S) transitions, using 158×10^{6} ϒ(2S) events accumulated with the Belle detector at the KEKB asymmetric energy electron-positron collider. No significant A^{0} signals in the mass range from the τ^{+}τ^{-} or μ^{+}μ^{-} threshold to 9.2  GeV/c^{2} are observed. We set the upper limits at 90% credibility level (C.L.) on the product branching fractions for ϒ(1S)→γA^{0} and A^{0}→τ^{+}τ^{-} varying from 3.8×10^{-6} to 1.5×10^{-4}. Our results represent an approximately twofold improvement on the current world best upper limits for the ϒ(1S)→γA^{0}(→τ^{+}τ^{-}) production. For A^{0}→μ^{+}μ^{-}, the upper limits on the product branching fractions for ϒ(1S)→γA^{0} and A^{0}→μ^{+}μ^{-} are at the same level as the world average limits, and vary from 3.1×10^{-7} to 1.6×10^{-5}. The upper limits at 90% credibility level on the Yukawa coupling f_{ϒ(1S)} and mixing angle sinθ_{A^{0}} are also given.

Punzi-loss:: a non-differentiable metric approximation for sensitivity optimisation in the search for new particles

We present the novel implementation of a non-differentiable metric approximation and a corresponding loss-scheduling aimed at the search for new particles of unknown mass in high energy physics experiments. We call the loss-scheduling, based on the minimisation of a figure-of-merit related function typical of particle physics, a Punzi-loss function, and the neural network that utilises this loss function a Punzi-net. We show that the Punzi-net outperforms standard multivariate analysis techniques and generalises well to mass hypotheses for which it was not trained. This is achieved by training a single classifier that provides a coherent and optimal classification of all signal hypotheses over the whole search space. Our result constitutes a complementary approach to fully differentiable analyses in particle physics. We implemented this work using PyTorch and provide users full access to a public repository containing all the codes and a training example.

Measurement of Differential Branching Fractions of Inclusive B→X_{u}ℓ^{+}ν_{ℓ} Decays

The first measurements of differential branching fractions of inclusive semileptonic B→X_{u}ℓ^{+}ν_{ℓ} decays are performed using the full Belle data set of 711  fb^{-1} of integrated luminosity at the ϒ(4S) resonance and for ℓ=e, μ. With the availability of these measurements, new avenues for future shape-function model-independent determinations of the Cabibbo-Kobayashi-Maskawa matrix element |V_{ub}| can be pursued to gain new insights in the existing tension with respect to exclusive determinations. The differential branching fractions are reported as a function of the lepton energy, the four-momentum-transfer squared, light-cone momenta, the hadronic mass, and the hadronic mass squared. They are obtained by subtracting the backgrounds from semileptonic B→X_{c}ℓ^{+}ν_{ℓ} decays and other processes, and corrected for resolution and acceptance effects.

Test of Lepton-Flavor Universality in B→K^{*}ℓ^{+}ℓ^{-} Decays at Belle

We present a measurement of R_{K^{*}}, the branching fraction ratio B(B→K^{*}μ^{+}μ^{-})/B(B→K^{*}e^{+}e^{-}), for both charged and neutral B mesons. The ratio for the charged case R_{K^{*+}} is the first measurement ever performed. In addition, we report absolute branching fractions for the individual modes in bins of the squared dilepton invariant mass q^{2}. The analysis is based on a data sample of 711  fb^{-1}, containing 772×10^{6}  BB[over ¯] events, recorded at the ϒ(4S) resonance with the Belle detector at the KEKB asymmetric-energy e^{+}e^{-} collider. The obtained results are consistent with standard model expectations.

Moving away from the "unit cost". Predicting country-specific average cost curves of VMMC services accounting for variations in service delivery platforms in sub-Saharan Africa

BACKGROUND

One critical element to optimize funding decisions involves the cost and efficiency implications of implementing alternative program components and configurations. Program planners, policy makers and funders alike are in need of relevant, strategic data and analyses to help them plan and implement effective and efficient programs. Contrary to widely accepted conceptions in both policy and academic arenas, average costs per service (so-called "unit costs") vary considerably across implementation settings and facilities. The objective of this work is twofold: 1) to estimate the variation of VMMC unit costs across service delivery platforms (SDP) in Sub-Saharan countries, and 2) to develop and validate a strategy to extrapolate unit costs to settings for which no data exists.

METHODS

We identified high-quality VMMC cost studies through a literature review. Authors were contacted to request the facility-level datasets (primary data) underlying their results. We standardized the disparate datasets into an aggregated database which included 228 facilities in eight countries. We estimated multivariate models to assess the correlation between VMMC unit costs and scale, while simultaneously accounting for the influence of the SDP (which we defined as all possible combinations of type of facility, ownership, urbanicity, and country), on the unit cost variation. We defined SDP as any combination of such four characteristics. Finally, we extrapolated VMMC unit costs for all SDPs in 13 countries, including those not contained in our dataset.

RESULTS

The average unit cost was 73 USD (IQR: 28.3, 100.7). South Africa showed the highest within-country cost variation, as well as the highest mean unit cost (135 USD). Uganda and Namibia had minimal within-country cost variation, and Uganda had the lowest mean VMMC unit cost (22 USD). Our results showed evidence consistent with economies of scale. Private ownership and Hospitals were significant determinants of higher unit costs. By identifying key cost drivers, including country- and facility-level characteristics, as well as the effects of scale we developed econometric models to estimate unit cost curves for VMMC services in a variety of clinical and geographical settings.

CONCLUSION

While our study did not produce new empirical data, our results did increase by a tenfold the availability of unit costs estimates for 128 SDPs in 14 priority countries for VMMC. It is to our knowledge, the most comprehensive analysis of VMMC unit costs to date. Furthermore, we provide a proof of concept of the ability to generate predictive cost estimates for settings where empirical data does not exist.

Evidence for X(3872)→J/ψπ^{+}π^{-} Produced in Single-Tag Two-Photon Interactions

We report the first evidence for X(3872) production in two-photon interactions by tagging either the electron or the positron in the final state, exploring the highly virtual photon region. The search is performed in e^{+}e^{-}→e^{+}e^{-}J/ψπ^{+}π^{-}, using 825  fb^{-1} of data collected by the Belle detector operated at the KEKB e^{+}e^{-} collider. We observe three X(3872) candidates, where the expected background is 0.11±0.10 events, with a significance of 3.2σ. We obtain an estimated value for Γ[over ˜]_{γγ}B(X(3872)→J/ψπ^{+}π^{-}) assuming the Q^{2} dependence predicted by a cc[over ¯] meson model, where -Q^{2} is the invariant mass squared of the virtual photon. No X(3915)→J/ψπ^{+}π^{-} candidates are found.

Selective background Monte Carlo simulation at Belle II

The large volume of data expected to be produced by the Belle II experiment presents the opportunity for studies of rare, previously inaccessible processes. Investigating such rare processes in a high data volume environment necessitates a correspondingly high volume of Monte Carlo simulations to prepare analyses and gain a deep understanding of the contributing physics processes to each individual study. This resulting challenge, in terms of computing resource requirements, calls for more intelligent methods of simulation, in particular for processes with very high background rejection rates. This work presents a method of predicting in the early stages of the simulation process the likelihood of relevancy of an individual event to the target study using graph neural networks. The results show a robust training that is integrated natively into the existing Belle II analysis software framework.

Search for an Invisibly Decaying Z^{'} Boson at Belle II in e^{+}e^{-}→μ^{+}μ^{-}(e^{±}μ^{∓}) Plus Missing Energy Final States

Theories beyond the standard model often predict the existence of an additional neutral boson, the Z^{'}. Using data collected by the Belle II experiment during 2018 at the SuperKEKB collider, we perform the first searches for the invisible decay of a Z^{'} in the process e^{+}e^{-}→μ^{+}μ^{-}Z^{'} and of a lepton-flavor-violating Z^{'} in e^{+}e^{-}→e^{±}μ^{∓}Z^{'}. We do not find any excess of events and set 90% credibility level upper limits on the cross sections of these processes. We translate the former, in the framework of an L_{μ}-L_{τ} theory, into upper limits on the Z^{'} coupling constant at the level of 5×10^{-2}-1 for M_{Z^{'}}≤6  GeV/c^{2}.

Search for B^{-}→μ^{-}ν[over ¯]_{μ} Decays at the Belle Experiment

We report the results of a search for the rare, purely leptonic decay B^{-}→μ^{-}ν[over ¯]_{μ} performed with a 711  fb^{-1} data sample that contains 772×10^{6}  BB[over ¯] pairs, collected near the ϒ(4S) resonance with the Belle detector at the KEKB asymmetric-energy e^{+}e^{-} collider. The signal events are selected based on the presence of a high momentum muon and the topology of the rest of the event showing properties of a generic B-meson decay, as well as the missing energy and momentum being consistent with the hypothesis of a neutrino from the signal decay. We find a 2.4 standard deviation excess above background including systematic uncertainties, which corresponds to a branching fraction of B(B^{-}→μ^{-}ν[over ¯]_{μ})=(6.46±2.22±1.60)×10^{-7} or a frequentist 90% confidence level interval on the B^{-}→μ^{-}ν[over ¯]_{μ} branching fraction of [2.9,10.7]×10^{-7}.

Search for CP Violation and Measurement of the Branching Fraction in the Decay D^{0}→K_{S}^{0}K_{S}^{0}

We report a study of the decay D^{0}→K_{S}^{0}K_{S}^{0} using 921  fb^{-1} of data collected at or near the ϒ(4S) and ϒ(5S) resonances with the Belle detector at the KEKB asymmetric energy e^{+}e^{-} collider. The measured time-integrated CP asymmetry is A_{CP}(D^{0}→K_{S}^{0}K_{S}^{0})=(-0.02±1.53±0.02±0.17)%, and the branching fraction is B(D^{0}→K_{S}^{0}K_{S}^{0})=(1.321±0.023±0.036±0.044)×10^{-4}, where the first uncertainty is statistical, the second is systematic, and the third is due to the normalization mode (D^{0}→K_{S}^{0}π^{0}). These results are significantly more precise than previous measurements available for this mode. The A_{CP} measurement is consistent with the standard model expectation.

Disrupting end-of-life cancer care delivery: Results from the engagment of patients with advanced cancer trial

6525

Background: Sustainable approaches to improve quality and safety of care of patients with advanced cancer while concurrently reducing costs is a growing national need. As part of the Veterans Administration Engagement of Patients with Advanced Cancer (EPAC) trial, we trained a lay health worker (LHW) to engage patients with stage 3 and 4 cancer in early advance care planning (ACP). The goal of this follow-up study was to examine the effect of the LHW intervention on patient-reported care experiences, healthcare utilization, and costs in the last 30 days of life for patients who died within 15 months of enrollment. Methods: We evaluated patient-reported experiences with decision-making, healthcare utilization, and total healthcare costs 30 days prior to death. A T-test was used to compare patient experiences with decision-making. To compare ED use and hospitalizations, we utilized an exact Poisson regression. A generalized linear model with gamma link-log function was used to compare total costs. The latter methods adjusted for length of follow-up. Results: In the 30 days prior to death, 60 patients died in each arm within 15 months of enrollment (difference not statistically significant). Patients in the intervention had significantly improved rates of ACP documentation (98% versus 18% p < 0.001), improved experiences with decision-making as measured by an index ranging from 0-5 with higher values representing more favorable experience (4.73 (SD 0.61) vs 4.15 (SD 1.02)) p < 0.001), higher utilization of hospice (77% vs 52%, p < 0.005), lower rates of any emergency department use (5% versus 45% p < 0.001) and any hospitalization (5% versus 43% p < 0.001), and significantly lower total costs of care ($1,048 versus $23,482 p < 0.001) compared to the patients randomized to the usual care arm. Conclusions: Integrating a LHW into oncology care to engage patients in early advance care planning resulted in significantly improved patient experience, decreased utilization and decreased total costs in the last month of life. LHWs may represent a sustainable resource to facilitate optimal patient-centered cancer care at the end-of-life. Clinical trial information: NCT02966509.

Chronic Postsurgical Pain Outcomes in Breast Reconstruction Patients Receiving Perioperative Transversus Abdominis Plane Catheters at the Donor Site: A Prospective Cohort Follow-up Study

BACKGROUND

Chronic postsurgical pain (CPSP) is a debilitating and costly condition. Risk factors for CPSP after autologous breast reconstruction have not been clearly established. Previously, we demonstrated that transversus abdominis plane (TAP) catheters delivering intermittent local anesthetic reduced postoperative morphine consumption. This prospective follow-up study aimed to (1) compare the incidence of CPSP after autologous breast reconstruction between patients who received postoperative intermittent TAP catheters with bupivacaine or saline boluses and (2) assess the factors that contribute to the development and maintenance of CPSP in this study cohort.

METHODS

Ninety-three patients who underwent deep inferior epigastric artery perforator or muscle-sparing transverse rectus abdominis breast reconstruction were randomized to receive TAP catheters with bupivacaine or saline postoperatively. Subsequently, patients were followed for a year to assess persistent pain, pain severity, quality of life scores, and functional disability at 6 and 12 months after surgery.

RESULTS

Twenty-four percent and 23% of patients reported CPSP at 6 and 12 months, respectively. There were no significant differences between groups (bupivacaine vs. placebo) on pain-related variables, including incidence of CPSP. Patients who reported greater variability in pain scores at rest over the first 48 hours postoperatively were more likely to have CPSP 6 months, but not 12 months, later.

CONCLUSIONS

Acute postoperative pain variability may contribute to the development of CPSP up to 6 months after autologous breast reconstruction surgery. Neither postoperative use of bupivacaine vs. saline in the TAP catheters nor acute pain severity influenced the 6- or 12-month incidence of CPSP.

Dose reduction in paediatric cranial CT via iterative reconstruction: a clinical study in 78 patients

AIM

To assess how adaptive statistical iterative reconstruction (ASIR) contributes to dose reduction and affects image quality of non-contrast cranial computed tomography (cCT) in children.

MATERIALS AND METHODS

Non-contrast cranial CT acquired in 78 paediatric patients (age 0-12 years) were evaluated. The images were acquired and processed using four different protocols: Group A (control): 120 kV, filtered back projection (FBP), n=18; Group B: 100 kV, FBP, n=22; Group C: 100 kV, scan and reconstruction performed with 20% ASIR, n=20; Group D1: 100 kV, scan and reconstruction performed with 30% ASIR, n=18; Group D2: raw data from Group D1 reconstructed using a blending of 40% ASIR and 60% FBP, n=18. The effective dose was calculated and the image quality was assessed quantitatively and qualitatively.

RESULTS

Compared to Group A, Groups C and D1/D2 showed a significant reduction of the dose-length product (DLP) by 34.4% and 64.4%, respectively. All experimental groups also showed significantly reduced qualitative levels of noise, contrast, and overall diagnosability. Diagnosis-related confidence grading showed Group C to be adequate for everyday clinical practice. Quantitative measures of Groups B and C were comparable to Group A with only few parameters compromised. Quantitative scores in Groups D1 and D2 were mainly lower compared to Group A, with Group D2 performing better than Group D1. Group D2 was considered adequate for follow-up imaging of severe acute events such as bleeding or hydrocephalus.

DISCUSSION

The use of ASIR combined with low tube voltage may reduce radiation significantly while maintaining adequate image quality in non-contrast paediatric cCT.

Effects of a multicomponent intervention to streamline initiation of antiretroviral therapy in Africa: a stepped-wedge cluster-randomised trial

BACKGROUND

In Africa, up to 30% of HIV-infected patients who are clinically eligible for antiretroviral therapy (ART) do not start timely treatment. We assessed the effects of an intervention targeting prevalent health systems barriers to ART initiation on timing and completeness of treatment initiation.

METHODS

In this stepped-wedge, non-blinded, cluster-randomised controlled trial, 20 clinics in southwestern Uganda were randomly assigned in groups of five clinics every 6 months to the intervention by a computerised random number generator. This procedure continued until all clinics had crossed over from control (standard of care) to the intervention, which consisted of opinion-leader-led training and coaching of front-line health workers, a point-of-care CD4 cell count testing platform, a revised counselling approach without mandatory multiple pre-initiation sessions, and feedback to the facilities on their ART initiation rates and how they compared with other facilities. Treatment-naive, HIV-infected adults (aged ≥18 years) who were clinically eligible for ART during the study period were included in the study population. The primary outcome was ART initiation 14 days after first clinical eligibility for ART. This study is registered with ClinicalTrials.gov, number NCT01810289.

FINDINGS

Between April 11, 2013, and Feb 2, 2015, 12 024 eligible patients visited one of the 20 participating clinics. Median CD4 count was 310 cells per μL (IQR 179-424). 3753 of 4747 patients (weighted proportion 80%) in the intervention group had started ART by 2 weeks after eligibility compared with 2585 of 7066 patients (38%) in the control group (risk difference 41·9%, 95% CI 40·1-43·8). Vital status was ascertained in a random sample of 208 patients in the intervention group and 199 patients in the control group. Four deaths (2%) occurred in the intervention group and five (3%) occurred in the control group.

INTERPRETATION

A multicomponent intervention targeting health-care worker behaviour increased the probability of ART initiation 14 days after eligibility. This intervention consists of widely accessible components and has been tested in a real-world setting, and is therefore well positioned for use at scale.

FUNDING

National Institute of Allergy and Infectious Diseases (NIAID) and the President's Emergency Fund for AIDS Relief (PEPFAR).

Acceptance of Advance Directives and Palliative Care Referral for Veterans With Advanced Cancer

Objectives:

To evaluate the documentation of advance directive (ADs) and physician orders for life-sustaining treatment (POLST) with acceptance of palliative care (PC) services referral among patients with cancer.

Methods:

We retrospectively reviewed veterans with advanced cancers at the Veterans Administration Palo Alto Health Care System. Chi-square tests estimated AD and POLST documentation and referral to PC. Logistic regression models estimated the odds of AD and POLST documentation and PC referral.

Results:

Two hundred and forty-six veterans were diagnosed with cancer. In all, 53% had a documented AD, 5% had a POLST, and 47% accepted referral to PC. The AD documentation was not associated with acceptance of PC.

Discussion:

We found no association of AD documentation and PC referral. Future studies should evaluate other factors that influence referral to these services.

Radiation dose reduction in CT with adaptive statistical iterative reconstruction (ASIR) for patients with bronchial carcinoma and intrapulmonary metastases

AIM

To compare the radiation dose and image quality of 64-row chest computed tomography (CT) in patients with bronchial carcinoma or intrapulmonary metastases using full-dose CT reconstructed with filtered back projection (FBP) at baseline and reduced dose with 40% adaptive statistical iterative reconstruction (ASIR) at follow-up.

MATERIALS AND METHODS

The chest CT images of patients who underwent FBP and ASIR studies were reviewed. Dose-length products (DLP), effective dose, and size-specific dose estimates (SSDEs) were obtained. Image quality was analysed quantitatively by signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) measurement. In addition, image quality was assessed by two blinded radiologists evaluating images for noise, contrast, artefacts, visibility of small structures, and diagnostic acceptability using a five-point scale.

RESULTS

The ASIR studies showed 36% reduction in effective dose compared with the FBP studies. The qualitative and quantitative image quality was good to excellent in both protocols, without significant differences. There were also no significant differences for SNR except for the SNR of lung surrounding the tumour (FBP: 35±17, ASIR: 39±22).

DISCUSSION

A protocol with 40% ASIR can provide approximately 36% dose reduction in chest CT of patients with bronchial carcinoma or intrapulmonary metastases while maintaining excellent image quality.

Reducing Radiation Dose in Adult Head CT using Iterative Reconstruction - A Clinical Study in 177 Patients

PURPOSE

To assess how ASIR (adaptive statistical iterative reconstruction) contributes to dose reduction and affects image quality of non-contrast cranial computed tomography (cCT).

MATERIALS AND METHODS

Non-contrast emergency CT scans of the head acquired in 177 patients were evaluated. The scans were acquired and processed using four different protocols: Group A (control): 120 kV, FBP (filtered back projection) n = 71; group B1: 120 kV, scan and reconstruction performed with 20 % ASIR (blending of 20 % ASIR and 80 % FBP), n = 86; group B2: raw data from group B1 reconstructed using a blending of 40 % ASIR and 60 % FBP, n = 74; group C1: 120 kV, scan and reconstruction performed with 30 % ASIR, n = 20; group C2: raw data from group C1 reconstructed using a blending of 50 % ASIR and 50 % FBP, n = 20. The effective dose was calculated. Image quality was assessed quantitatively and qualitatively.

RESULTS

Compared to group A, groups B1/2 and C1/2 showed a significantly reduced effective dose of 40.4 % and 73.3 % (p < 0.0001), respectively. Group B1 and group C1/2 also showed significantly reduced quantitative and qualitative image quality parameters. In group B2, quantitative measures were comparable to group A, and qualitative scores were lower compared to group A but higher compared to group B1. Diagnostic confidence grading showed groups B1/2 to be adequate for everyday clinical practice. Group C2 was considered acceptable for follow-up imaging of severe acute events such as bleeding or subacute stroke.

CONCLUSION

Use of ASIR makes it possible to reduce radiation significantly while maintaining adequate image quality in non-contrast head CT, which may be particularly useful for younger patients in an emergency setting and in follow-up.

KEY POINTS

ASIR may reduce radiation significantly while maintaining adequate image quality. cCT protocol with 20 % ASIR and 40 %ASIR/60 %FBP blending is adequate for everyday clinical use. cCT protocol with 30 % ASIR and 50 %ASIR/50 %FBP blending is adequate for follow-up imaging

GM-CSF and ipilimumab therapy in metastatic melanoma: Clinical outcomes and immunologic responses

We conducted a phase II clinical trial of anti-CTLA-4 antibody (ipilimumab) and granulocyte-macrophage colony-stimulating factor (GM-CSF) in 22 patients with metastatic melanoma and determined clinical outcomes and immunologic responses. The treatment consisted of a 3-mo induction with ipilimumab at 10 mg/kg administered every 3 weeks for four doses in combination with GM-CSF at 125 µg/m² for 14 d beginning on the day of the ipilimumab infusion and then GM-CSF for 3 mo on the same schedule without ipilimumab. This was followed by maintenance therapy with the combination every 3 mo for up to 2 y or until disease progression or unacceptable toxicity. Blood samples for determination of immune subsets were obtained before treatment, at week 3 (end of cycle 1) and at week 6 (end of cycle 2). Blood samples were also obtained from seven subjects who were cancer-free. The immune response disease control (irDC) rate at 24 weeks was 41% and the overall response rate (ORR) was 32%. The median progression free-survival (PFS) was 3.5 mo and the median overall survival (OS) was 21.1 mo. 41% of the patients experienced Grade 3 to 4 adverse events. We conclude that this combination is safe and the results suggest the combination may be more effective than ipilimumab monotherapy. Further, the results suggest that lower levels of CD4⁺ effector T cells but higher levels of CD8⁺ T cells expressing PD-1 at pre-treatment could be a potential biomarker for disease control in patients who receive immunotherapy with ipilimumab and GM-CSF. Further trials of this combination are warranted.