FIRST in Cutaneous Squamous Cell Carcinoma
2026-08-14
SoCO Journal Club · August 14, 2026
Thanks for showing up.
Thanks to everyone who joined us on August 14. Here’s a look back at the paper, the questions that drove the discussion, how people participated, and who was in the room.
Meeting pulse
A good turnout — and people stuck around
People who joined
26
Unique cleaned names in the Teams attendance record
Median time together
87 min
Half the room stayed at least this long
Stayed at least an hour
73%
Joined for at least 60 minutes
Participation
A few ways people participated
Teams records a few simple signals — time in the meeting, unmuting, hand raises, and camera use. They do not tell the whole story, but they give us a quick sense of how people took part.
Stayed ≥ 60 minutes
73%
Camera on
73%
Unmuted
69%
Raised a hand
8%
Recorded Teams signals; these do not measure the quality or depth of participation.
The paper we discussed
FIRST in cutaneous squamous cell carcinoma
Primary article · Journal for ImmunoTherapy of Cancer
Frontline immunotherapy with response-guided subsequent treatment (FIRST) in cutaneous squamous cell carcinoma: a Bayesian causal analysis of dose intensity, early benefit, and treatment de-escalation
Miller DM, et al. J Immunother Cancer. 2026;14(7):e015029.
doi: 10.1136/jitc-2026-015029
From a real-world cohort to an inference question
Because treatment evolved over time, patients received different amounts of immunotherapy. That gave us an opportunity to first describe what happened across 189 patients, then ask the harder question: what does another dose actually buy us?
From cohort to inference
The analysis moved from describing what happened to asking what additional treatment may have contributed.
01 · Observe
189-patient real-world cohort
Patients received different amounts of immunotherapy during an evolving period of practice.
→
02 · Describe
What happened?
Characterize dose exposure, response, surgery, recurrence, and event-free survival.
→
03 · Infer
What might another dose buy us?
Use causal assumptions and Bayesian models to estimate dose–response and dose–EFS relationships.
189
patients
Dose → Response
one inferential target
Dose → EFS
second inferential target
The study moved from describing the observed cohort to estimating the relationship between dose and response and between dose and event-free survival. Bayesian models allowed uncertainty to be carried forward rather than reducing the analysis to a single point estimate.
An important caveat: this was an observational experience during an evolving period of practice—not a randomized comparison. Treatment selection and confounding by indication matter, and the results should be interpreted in that context.
Discussion highlights
What stuck with us from the conversation
The paper was the starting point, but a lot of the discussion was really about how we treat these patients now, how we decide when enough treatment may be enough, and how we make causal claims from observational data.
01 · The landscape
Practice is changing faster than consensus
Vishal opened by placing FIRST in the rapidly evolving treatment landscape for high-risk resectable CSCC. Christine and Howard then walked through the study schema and descriptive results, including the heterogeneity in real-world treatment exposure. That set up one of the central questions for discussion: with substantial variation in the number of doses patients received, what can these observational data actually tell us about how much treatment is necessary?
02 · The clinical question
Once a patient responds, how much more treatment is actually necessary?
Our current response-guided approach generally starts with early immunotherapy and reassessment. Patients with a deep clinical response may be observed without additional immunotherapy, surgery, or radiation; patients without that response retain additional treatment options. The clinical appeal is obvious. The harder question is what level of evidence should be required before treatment is safely omitted.
03 · Causal inference
The DAG made the assumptions visible
Isaac walked us through the directed acyclic graph for the dose–response question. Rather than simply adding another statistical graphic, the DAG forces us to make our causal assumptions explicit: what might confound the relationship, and what downstream variables should we avoid automatically adjusting for?
04 · Clinicians + statisticians
Are we already doing this — and do we really need a DAG?
Paul raised an important practical question: aren’t we already thinking about these relationships when we build our models, and what is the added value of formalizing them in a DAG? Dan and Isaac highlighted why formalization can be useful. Defining how disease, treatment decisions, patient characteristics, and outcomes relate to one another requires clinical domain expertise, while the DAG makes those assumptions explicit and open to scrutiny.
If a paper is making a causal claim from observational data, ask to see the causal assumptions—not just the regression model.
The group also talked about how unfamiliar DAGs still are in much of clinical research. Ade shared the perspective of having encountered this approach in editorial work—albeit rarely—while for many others it was relatively new. That led to a broader challenge for authors and reviewers: make causal assumptions explicit, discuss them with the clinical and statistical teams early, and expect more than a list of variables entered into a multivariable model.
Where the discussion landed
There was real enthusiasm for the response-guided treatment concept and for getting these data in front of a broader audience. Vern Sondak helped lead that discussion, emphasizing the potential for this approach to be genuinely paradigm-changing if the signal holds up prospectively.
At the same time, the group kept the evidentiary framing appropriately careful: this may represent an important shift in how we think about treatment intensity, but the observational findings still need prospective confirmation.
That combination of enthusiasm and caution felt like the right place to land: the signal is compelling, the clinical implications could be substantial, and the next step is to test the idea prospectively.
Our community
Who joined us?
Below is the full Teams roster. Where we could match someone to the SoCO community survey, we also included their role, institution, and how long they have been part of SoCO.
We appreciate the whole mix: people who had just recently found SoCO, those who have been with us for a while, and the originals who have been here since the beginning.
26 colleagues joined the meeting
Community details available for 25; “—” means not reported in the survey.
👋 First SoCO Meeting
| Name | Professional role | Institution |
|---|---|---|
| Elizabeth J. Lilley | Clinical Faculty / Practicing Clinician | Brigham and Women’s Hospital |
| Mina Bakhtiar | Clinical Faculty / Practicing Clinician | Massachusetts General Hospital |
| Tatyana Sharova | Other | MGH |
Less Than 1 Year
| Name | Professional role | Institution |
|---|---|---|
| Ade Adamson | Clinical Faculty / Practicing Clinician | University of Texas |
| Ajay N. Sharma | Clinical Faculty / Practicing Clinician | Massachusetts General Hospital |
| David J. Savage | Clinical Faculty / Practicing Clinician | University of New Mexico |
| Elizabeth I. Buchbinder | Clinical Faculty / Practicing Clinician | Mass General Brigham Cancer Institute |
| Mehran Behruj Yusuf | Clinical Faculty / Practicing Clinician | UAB |
| Rhoda Alani | Clinical Faculty / Practicing Clinician | MGH — Visiting Professor |
| Rima Kulikauskas | Other | University of Washington |
1–2 Years
| Name | Professional role | Institution |
|---|---|---|
| Dan Hippe | Other | Fred Hutch Cancer Center |
| Frances Collichio | Clinical Faculty / Practicing Clinician | University of North Carolina |
| Jennifer Desimone | Clinical Faculty / Practicing Clinician | INOVA Schar Cancer Institute |
| Peter Ch’en | Postgraduate Trainee — Resident or Fellow | University of Washington |
| Song Park | — | University of Washington |
More Than 2 Years
| Name | Professional role | Institution |
|---|---|---|
| Alex Sorrentino | Clinical Faculty / Practicing Clinician | Massachusetts General Hospital |
| Christine C. Cimoch | Advanced Practice Provider | MEEI |
| Jessica L. Fewkes | Clinical Faculty / Practicing Clinician | Mass Eye and Ear |
| Krista M. Rubin | Clinical Faculty / Practicing Clinician | Massachusetts General Hospital |
| Paul Nghiem | Clinical Faculty / Practicing Clinician | University of Washington |
| Samir Gupta | Clinical Faculty / Practicing Clinician | Mass Eye and Ear |
| Vern Sondak | Clinical Faculty / Practicing Clinician | Moffitt |
🏆 Since the Beginning — SoCO Originals
| Name | Professional role | Institution |
|---|---|---|
| David M. Miller | Clinical Faculty / Practicing Clinician | MGH |
| Howard L. Kaufman | Clinical Faculty / Practicing Clinician | MGB |
| Isaac Brownell | Clinical Faculty / Practicing Clinician | National Institute of Arthritis and Musculoskeletal and Skin Diseases Dermatology Branch |
| Vishal Patel | Clinical Faculty / Practicing Clinician | George Washington University |
About the roster. Everyone listed above appears in the Teams attendance record. Role, institution, and SoCO tenure come from the community survey when we could match the name. A dash means we did not have survey information to add.