BACKGROUND:Serious illness conversations (SICs) aim to elicit patient preferences and are associated with improved quality of life and reduced care utilization, but they occur infrequently. Sustainable interventions that encourage SICs are needed. PATIENTS AND METHODS:This pragmatic 4-arm randomized controlled trial enrolled adult patients at 5 disease-based oncology clinics at 2 sites of an academic cancer center between December 4, 2022, and July 31, 2024. All patients had pathways data indicating they were starting a treatment associated with a poor prognosis without documentation of an SIC in the Advance Care Planning module of the electronic health record (ACP-SICs) in the prior 6 months. Patients were randomized to 1 of 4 groups: (1) a nudge consisting of a mailed letter and questionnaire encouraging SICs; (2) a clinician nudge comprising an email reminder sent the day prior to the clinic visit to prompt an SIC; (3) both nudges; or (4) no nudges. The primary outcome was the proportion of patients with ACP-SICs within 60 days of randomization comparing the control (no-nudge) and combined-nudge arms; a prespecified alternate primary outcome included SICs identified in the free text of clinician notes using natural language processing (ACP + NLP-SIC). RESULTS:A total of 1,051 patients (median age, 65 years; 60% female; 79% White) were randomized to the control (n=261), clinician-nudge (n=240), patient-nudge (n=273), and combined-nudge arms (n=277). The ACP-SIC rates were 10.7%, 16.7%, 10.6%, and 17.3% for the control, clinician-nudge, patient-nudge, and combined-nudge arms, respectively, and the ACP + NLP-SIC rates were 22.6%, 28.8%, 22.3%, and 32.5%, respectively. Patients in the combined-nudge group had significantly higher ACP-SIC and ACP + NLP-SIC rates than the control group (P=.045 and P=.01, respectively), whereas the clinician-nudge and patient-nudge groups did not. CONCLUSIONS:Combined clinician- and patient-directed nudges resulted in higher SIC rates in 60 days, driven largely by the clinician nudge. NLP increased detection of SICs, demonstrating the importance of evaluating SICs in free-text notes.

PURPOSE Oncology treatment pathways provide decision support and encourage guideline adherence. Pathway data combined with electronic health record (EHR) data can identify patient populations with poor prognoses, low serious illness conversation (SIC) rates, and high acute care utilization that may benefit from targeted interventions. PATIENTS AND METHODS We conducted a retrospective cohort analysis among adults with cancer treated at seven affiliated sites of the Dana-Farber Cancer Institute (DFCI) who had navigations within 21 treatment pathways between July 29, 2019, and March 8, 2023. DFCI clinicians previously identified pathway nodes with an estimated survival less than 1 year, termed poor prognosis (PP) nodes. We combined pathway data with EHR data to calculate the median overall survival (OS) and proportion of patients with SICs, acute care utilization (hospitalizations and emergency department visits), and outpatient palliative care 6 months after treatment node navigation for all, PP, and nonpoor prognosis (nPP) nodes. SICs were identified using the EHR advanced care planning (ACP) tab. RESULTS There were 15,261 navigations for 10,203 patients (median age 66 years, 55% female, 85% White). The median OS was 13.8 months for all nodes, 7.8 months for PP nodes, and 21.0 months for nPP nodes. The ACP section of the EHR rate 6 months after navigation was 19.6% for PP nodes versus 11.0% for nPP nodes. There was substantial intragroup variability in OS and SIC rates among all nodes. SICs were recorded in the ACP tab for only 34.3% of decedents. Patients who navigated to PP nodes had higher levels of acute care utilization and palliative care encounters. CONCLUSION Treatment pathway data enabled identification of patient populations with poor prognoses, low SIC rates, and high acute care utilization.

12013 Background: Serious illness conversations (SICs) can improve quality of life and decrease intensive care utilization at the end of life for patients with cancer. Yet SIC rates for patients with cancer are low and sustainable strategies are needed to engage patients and oncology clinicians in SICs. Methods: This randomized controlled trial was conducted at a tertiary cancer center. Using a cancer treatment guideline program (Pathways), oncology subspecialists identified treatment decision points where patients have an average prognosis <1 year and they would recommend an SIC. We enrolled patients with breast, gastrointestinal, genitourinary, gynecologic, and thoracic cancers who reached these points and did not have SICs documented in the Advance Care Planning tab (ACP-SICs) of the electronic medical record in the previous 6 months. Patients were randomized to receive a patient nudge (a mailed letter encouraging discussion of their values and preferences with their oncologist; arm 1), a clinician nudge (emails sent to oncology clinicians encouraging an SIC the day prior to the clinic visit; arm 2), both nudges (arm 3), or no nudges (arm 4). The primary analysis compared ACP-SIC documentation 60 days post-randomization for the combined vs no-nudge arms (arm 3 vs 4) using a generalized estimating equation model with a logit link adjusted for disease center and prior clinician SIC training, clustered on oncologists. A pre-specified alternative primary outcome used natural language processing (NLP) to identify SICs in free text notes in the 6 months prior to randomization and 60 days after to evaluate the presence of an NLP- or ACP-SIC 60 days after randomization using the same model. Similarly-constructed Cox proportional hazards models were used to estimate time to SIC. Results: Among 1051 patients randomized (arm 1: 273, arm 2: 240, arm 3: 277, arm 4: 261), median age was 65 years (range: 25-94), 40% were male, 79% White, 52% had gastrointestinal and 20% breast cancers. The Table displays unadjusted ACP and NLP+SIC rates. In adjusted analyses, compared to patients in the no-nudge arm (arm 4), patients in the combined nudge arm (arm 3) had 79% higher odds of ACP-SIC at 60 days (odds ratio 1.79, 95% CI 1.11-2.88, p=0.02) and 59% higher odds of NLP+ACP-SIC at 60 days (odds ratio 1.59, 95% CI 1.14-2.22, p=0.006). Time to ACP-SIC was 59% faster in clinician nudge-containing arms than the no-nudge arm (adjusted HR 1.59, 95% CI 1.16-2.19, p=0.004). Conclusions: Clinician emails increased SICs within 60 days, whereas patient nudges were ineffective. NLP increased detection of SICs by 49.7%, demonstrating the importance of evaluating SICs in free-text notes in SIC interventions.Long-term analyses will evaluate the interventions’ impact on care delivery outcomes. Clinical trial information: NCT05629065 . Unadjusted SIC at 60 days. Arm 1: Patient 2: Clinician 3: Combined 4: None ACP, % 10.6 16.7 17.3 10.7 NLP+ACP, % 22.4 27.8 34.0 24.4

415 Background: Oral targeted therapies in lung cancer represent a costly but critical component of care. Currently, the approval process for these treatments can sometimes take several weeks. Assessing the approval rates and efficacy of on-pathway treatment recommendations is an important step towards bringing clinicians and payors together to streamline the decision support and authorization process. Methods: We retrospectively analyzed authorization requests for oral anti-cancer agents for lung cancer. Eligible requests were filed between August 1, 2022, to March 1, 2024, and represented new treatment starts for that agent. In addition to collecting information about final authorization decisions, we determined whether the use case was consistent with Dana-Farber Pathways treatment recommendations in place at the time of the order. Furthermore, we measured survival from the start of oral therapy. Median survivals were calculated using the Kaplan-Meier method. These were assessed for the whole cohort and for each class of genomic aberrations. Results: There were a total of 214 eligible oral therapy claims filed between August 1, 2022, to March 1, 2024. 147 (69%) of these were considered on pathway, while 67 (31%) were off pathway. In total, 100% of the on-pathway navigations received payor approval, compared to 93% of the off-pathway navigations. The estimated 12-month survival was 86.5% for the on-pathway group, compared to 56.5% for the off-pathway group. Median overall survival has not yet been reached for either group. EGFR and ALK targeted therapies accounted for the majority of authorization requests, with 114 and 28, respectively. Conclusions: With the use of Dana-Farber Lung Cancer Pathways, the majority of requests for targeted therapies were consistent with pathway recommendations. This on-pathway cohort not only achieved a 100% payor authorization rate, but the 12-month survival was superior to that of off-pathway requests. These data provide a meaningful starting point for pathways vendors and healthcare payors to develop mutually agreed upon recommendations for use of these costly but impactful therapies.

414 Background: Fewer than 10% of adult cancer patients participate in clinical trials. Clinical pathways platforms have traditionally sought to improve trial accrual by displaying available trials for eligible patients. Our team explored how Pathways navigation data can also be used to identify eligible patients of poorly accruing trials and provide study investigators with a near real-time data feed of those patients. Methods: Providers at Dana-Farber Cancer Institute are asked to navigate the pathways decision-support platform and order systemic cancer therapies 7 days in advance of administration to assist with authorization. We invited Principal Investigators (PIs) of poorly accruing trials to collaborate with our Pathways data team through our Pathways to Accrual to Clinical Trials (PACT) service. Through PACT, PIs are informed of patients who were navigated to nodes involving the trial in question but received standard therapy instead. PIs are given access to a Tableau Dashboard that is updated with new data daily and receive weekly reminders about the dashboard. The current analysis includes two trials that have been using PACT for at least one year. We reviewed monthly trial accrual rates from the time of the trials opening until March 2024. The dataset did not meet criteria for Poisson analysis, so Fischer’s exact test was used to compare accrual rates. We ensured the denominator of potentially eligible patients was relatively consistent. Results: DF-21-519 accrued 16 patients over 14 months prior to PACT and 33 patients over 12 months using PACT. DF-22-028 accrued 0 patients over 6 months prior to PACT and 12 patients over 14 months using PACT. Collectively, use of PACT more than doubled accrual rates for the involved trials (one-tailed p = 0.048). Conclusions: A real-time Pathways data feed can be used to identify patients for appropriate clinical resources. Here, we show how our PACT program helped to more than double clinical trial accrual for the involved studies. Scaling this technology across a broader swath of clinical trials could be an important advance for trial participation for users of this pathways platform. Trial Accruals, Pre PACT Accrual Rate, Pre PACT (Accruals/month) Accruals, Post PACT Accrual Rate, Post PACT (Accruals/month) DF-21-519 16 accruals in 14 mo 1.14 33 accruals in 12 mo 2.75 DF-22-028 0 accruals in 6 mo 0 12 accruals in 14 mo 0.86 Total 16 accruals in 20 mo 0.8 45 accruals in 26 mo 1.73 one-tailed p is 0.048

PURPOSE The NCI-MATCH trial assigned patients with solid tumors, lymphomas, or multiple myeloma to targeted therapies on the basis of identified genetic alterations from tumor biopsies. In preclinical models, neurofibromatosis 2 (NF2)-inactivated tumors display sensitivity to focal adhesion kinase (FAK) inhibition. The EAY131-U subprotocol evaluated the efficacy of defactinib, a FAK inhibitor, in patients with NF2-altered tumors. METHODS Patients whose tumors harbored an inactivating NF2 mutation on next-generation sequencing were assigned to subprotocol U. Defactinib 400 mg was given orally twice a day until progression or intolerable toxicity. The primary end point was objective response rate (ORR), secondary end points included toxicity, progression-free survival (PFS), and 6-month PFS. RESULTS Of 5,548 patients with sufficient tissue for genomic analysis, 57 patients were found to have NF2 alterations. Thirty-five patients ultimately enrolled and 33 were treated, with one not having central confirmation and two ineligible for outcome analysis. All patients had received previous treatment, with 52% having received three or more previous lines of therapy. The most common treatment-related toxicities were fatigue (36%), nausea (33%), and hyperbilirubinemia (27%), with 27% of patients having grade 3 toxicities. Median follow-up was 35.9 months with an ORR of 3% from one partial response in a patient with choroid meningioma. Among the 12 patients (40%) with a best response of stable disease, eight demonstrated some tumor shrinkage. Median PFS was 1.9 months, and six patients achieved a PFS >5.5 months. No correlation was identified between clinical outcomes and tumor histology or specific NF2 genotype. CONCLUSION This protocol did not meet its prespecified primary end point. Defactinib monotherapy had limited clinical activity in this cohort of previously treated patients with solid tumors exhibiting NF2 loss.

12027 Background: Treatment pathways are widely used to provide decision support and encourage guideline adherence. Pathways data may be used to identify patients for targeted interventions to improve cancer care. For example, pathways data may enable the identification of patients with poor prognoses for whom serious illness conversations (SICs) should be prioritized to ensure goal-concordant care. Methods: For patients starting new therapies at Dana-Farber, oncologists select a node in clinical pathways to indicate the line of treatment. Subspecialty oncologists identified “poor prognosis” nodes in pathways, defined as therapies for patients with expected survivals of < 12 months and for whom an SIC would be appropriate (e.g., 3 rd -line treatment for metastatic colon cancer). Pathway node navigations for patients with metastatic solid tumor malignancies with poor prognosis nodes (n = 21 disease pathways) were combined with electronic medical record data to identify SIC documentation in an advance care planning module within 6 months of pathway navigation and patients’ dates of death if present. For each node, we calculated the median overall survival (OS), the proportion of patients with an SIC, and the proportion of patients who died before reaching the next node. Results: There were 10,132 navigations for 7,031 patients (median age 67 years, 52% female, 86% White) between 8/16/2019 and 1/4/2023. With each treatment line, patients’ median OS decreased and SIC rates increased (examples for 3 diseases in Table). Among patients who reached poor prognosis nodes, the median OS was 4.9 months, mean SIC rate was 43.8%, and 60.2% of patients died without reaching a subsequent node. For nodes immediately prior to poor prognosis nodes, the median OS was 7.1 months, the mean SIC rate was 25.2%, and 31.3% of patients died without reaching the next node. 46.3% of patients died without reaching a poor prognosis node. The proportion of patients having an SIC at death was 40.9% among all decedents, 50.8% for decedents reaching a poor prognosis node, and 29.3% for decedents who do not reach a poor prognosis node. Conclusions: Clinical pathways can be used as a scalable method to identify patients with metastatic solid tumors and poor prognoses. In our study, only half of patients who reached a poor prognosis node had an SIC before death and nearly half of decedents never reached a poor prognosis node, underscoring the importance of identifying this population to improve end-of-life care. A clinical trial using pathways to identify and deliver interventions to increase SICs is ongoing. [Table: see text]

<p>Supplementary Table S3. List of LKB1 mutant lung adenocarcinoma specimens by OncoMap tested for LKB1 IHC.</p>

Supplementary Table S2. Panel of cancer cell lines with known LKB1 status tested for LKB1 IHC.

Supplemental Figure 2. A, KRAS G12C mutation is the most frequently mutated in smokers while G12D is most common in never-smokers. B and C, Never-smokers were significantly more likely than former or current smokers to have a transition mutation (G>A) rather than transversion mutations known to be smoking-related (G>T or G>C).

<p>Table S4. LKB1 patient characteristics in KRAS mutant patients (n=148)*</p>

Shown is the schema of study design and enrollment.(30) As noted, the cut-off for data analysis was May 15, 2015.

<p>Supplementary Table S1. KRAS mutant NSCLC patients' characteristics (n=514)</p>

<p>Table S4. LKB1 patient characteristics in KRAS mutant patients (n=148)*</p>

XLSX file - 25K, cfDNA results for figures 2, 4, 5, calculated both as concentration per mL plasma and as percent mutant.

<p>Supplemental Figure 1. A, OS in stage IV KRAS-mutant NSCLC according to KRAS tranversion or transition mutations and LKB1 status. B, OS by different KRAS mutations (G12C vs. G12D)</p>

<p>Supplementary Table S1. KRAS mutant NSCLC patients' characteristics (n=514)</p>

e13523 Background: Dana-Farber Cancer Institute’s (DFCI) clinical pathways provide expert oncology decision-support. We schedule 2-4 reviews/year for each of over 40 medical oncology pathways, plus ad hoc reviews. When FDA issues new approvals or label expansions, the relevant medical director assigns it to the next scheduled review or a more urgent ad hoc meeting. Here we explore the time to initial review (TTIR) for FDA approvals. Methods: This analysis includes FDA approvals/label expansions for adult oncology patients from 01/01/2021 – 12/31/2022. We excluded CAR-T therapies as well as FDA approvals not directed at one of our existing pathways. TTIR was calculated from date of FDA approval to date of DFCI clinical pathways review. Univariate analyses examined the impact of prior in-class competitors in the specific approval setting; prior approvals for the treatment class in general; and scheduled meeting cadence. Results: From January 1, 2021 – December 31, 2022, there were 58 FDA approvals/label expansions in diseases for which a DFCI pathway exists. Eighteen represented items that had been previously addressed (e.g., conversion of accelerated approval to full, or a label expansion adopted into pathways earlier based on published data). For the remaining 40 new approvals or label expansions, the median TTIR was 3.4 months. 43% of FDA approvals were addressed prior to or within 30 days after announcement. There were 9 ad hoc reviews over this period, including review of 3 new FDA approvals. Median TTIR for items that were a first-in-class therapy for that specific indication (n= 20) was 2.1 months, compared to 3.5 months for agents with an existing in-class competitor in that setting (n=20). Median TTIR for items that represented the first approval for a novel first-in-class agent (n=13) was 1.5 months; TTIR for agents with existing in-class approvals in any setting (n=27) was 3.5 months. Median TTIR for FDA actions involving diseases/pathways that have > 3 scheduled review meetings/yr (n=28) was 2.8 months, compared to 5.2 months in pathways with 1-2 scheduled reviews/yr (n=12). Conclusions: Clinical Pathways provide a structured mechanism for reviewing whether, when, and how new treatments are best used in oncology practice, and for disseminating those recommendations to providers. The time to initial review was shorter for initial registrational approvals for novel agents, and for treatments in diseases with more frequent meetings. [Table: see text]

492 Background: Dana-Farber Cancer Institute’s (DFCI) Clinical Pathways provide evidence-based decision support for oncologists across a variety of care settings. Data generated by Pathways enable continued cancer care improvement and may also be a tool for identifying patient candidates for survivorship. Here, we explore patients diagnosed with testicular cancer treated with curative intent. Methods: Oncologists at DFCI navigate Clinical Pathways when starting or changing therapies. Each navigation provides discrete clinical data elements about disease (e.g., histology, stage), treatment setting (e.g., line of therapy), and treatment selected. Here, pathway navigation data was used to retrospectively identify patients with testicular cancer who were appropriate for survivorship resources. Eligibility criteria included patients who completed therapy, had not received treatment in 12-24 months since initial therapy navigation, and had a follow-up appointment scheduled. Eligible patients were referred to the testicular cancer survivorship program. Results: Between April 2021 to April 2022, we identified 25 testicular cancer patients treated at DFCI’s primary Longwood campus. Twelve of the 25 patients were identified as eligible for a survivorship referral. Nine of those 12 patients (75%) were successfully identified through the Pathways testicular survival algorithm described above. A Pathways navigation had not been completed by the providers of the 3-remaining survivorship-eligible patients, who were instead identified by EHR treatment plan data. One of the 12 patients screened by the testicular survivorship identification process developed metastatic disease that was not captured by the current algorithm and was then deemed inappropriate for survivorship referral. The remaining 11 patients were contacted and offered survivorship resources. As a reference, the total pathways usage rate between April 2021 to April 2022 for the DFCI genitourinary oncology group was 85% across all GU malignancies, and 74% in testicular cancer. Conclusions: Pathways Navigations can provide robust clinical information about patients, disease, and treatment. This information can be used to identify patients appropriate for survivorship and other clinical resources and trial opportunities in testicular cancer and across other malignancies. The impact of this process requires reliable provider usage of the Clinical Pathways tool.[Table: see text]

1589 Background: Cancer patients in the last year of life have different clinical needs and evolving goals of care. Using our oncology decision-support pathways to help clinicians consistently identify such patients in a systematic and prospective fashion, at a discrete moment in the care trajectory, may be an important step towards matching the care of these patients with their stated goals. Methods: Medical oncologists from each disease group at the Dana-Farber Cancer Institute (DFCI) were tasked with identifying clinical settings in each oncology care pathway associated with an expected median survival of < 12 months. This information was embedded into the underlying data model of the pathways platform, allowing us to determine how often clinicians navigated through each poor prognosis node. Results: From 3/1/20 – 6/30/21, there were 264 navigations in 205 unique lung cancer patients receiving standard of care (i.e., not on clinical trial) for a clinical condition associated with poor prognosis. Overall, the median overall survival from the time of a patient’s first navigation through a poor prognosis node during the defined study period was 6.4 months. Table lists outcomes for each specific setting. Patients with squamous or small cell lung cancer being treated in or beyond the third-line setting had notably poor outcomes, with less than a third of these patients surviving 6 months from the time of navigation. Conclusions: A clinical pathways platform can be a key tool in designating clinical scenarios associated with poor prognosis and identifying patients who may be particularly at risk. Pathways analytics provide real-world evidence corroborating the expected poor prognosis based on published studies and can identify specific clinical subsets for whom specific resources are warranted. By embedding this into the pathways data model, we aim to alert physicians to conduct goals of care conversations, offer supportive care resources, and match patients to appropriate treatment options and clinical trials. [Table: see text]