A case for stronger governance when off-label salvage therapy requests collide with plan integrity, evidence boundaries, and executive leverage

Note: Details are anonymized and blended from recurring real-world patterns to focus on governance—not any one case.

The Call That Bends the System

The call came on a Thursday afternoon. Not to the medical director—to the VP of Sales. A large employer group had a member on QINLOCK for advanced GIST. The coverage decision was wrong, apparently. Someone wanted it “handled.”

QINLOCK is FDA-approved for a very specific patient profile: advanced, unresectable, or metastatic gastrointestinal stromal tumor after prior treatment with three or more kinase inhibitors. It’s a late-line option with a constrained evidence base.

The labeled dose is 150 mg once daily—three 50 mg tablets. There’s clinical experience around escalation to 300 mg after progression, and we’d built a formal exception pathway for that.

What came through wasn’t 300 mg. It was 450 mg—nine tablets a day, triple the labeled dose, with thin-to-nonexistent evidence supporting it.

On a claim, it still looks like QINLOCK. The NDC is right. The quantity could be explained as escalation if you squint. But the clinical rationale had left the building. What remained was pressure.

The implied threat was never stated directly. It didn’t need to be. Everyone understood: accommodate this, or the account walks.

This wasn’t a clinical question anymore. It was a power question. And how you answer power questions—when the money is real and the patient is dying—reveals what kind of system you’ve actually built.

What Stewardship Actually Means

Stewardship gets called rationing by people who want exceptions. It’s not.

A health plan manages pooled premiums. Every dollar spent on one member is unavailable for another. That’s not ideology—it’s arithmetic. The ethical obligation isn’t to say yes to everything. It’s to be consistent, transparent, and fair about how yes and no are reached.

The rule is simple: materially similar situations should get materially similar answers—unless differences are documented and defensible.

When a coverage decision changes based on who’s asking, you don’t just risk a wrong call. You weaken the system’s legitimacy, because the reasons are no longer clinical.

The moral burden isn’t just the decision. It’s the precedent. Every exception granted under pressure becomes a reference point for the next request. Until you don’t have a policy anymore—you have a negotiation.

Why Oncology Exceptions Are Uniquely Fragile

Oncology is where governance gets stress-tested.

The clinical context is brutal: time is short, families are desperate, and physicians advocate hard for anything that might help. Nobody wants to be the person who said no to a dying patient.

But that emotional weight is exactly why governance matters more here, not less.

In salvage oncology, evidence is often extrapolated, underpowered, or based on surrogate endpoints. QINLOCK’s pivotal trial—INVICTUS—showed a median progression-free survival advantage of about 6.3 months versus about 1 month on placebo. That sounds dramatic until you sit with what it means: half of patients see less than that before the cancer advances anyway. You’re not buying remission. You’re buying time—maybe—measured in months.

Whether that tradeoff is “worth it” isn’t a question I can answer for someone facing a terminal diagnosis. But it is a question the plan has to answer consistently, regardless of who is asking.

When evidence is thin and emotions are high, the risk isn’t only clinical uncertainty. It’s governance drift: one-off approvals that never get written down, standards that bend depending on influence, precedents nobody formally endorsed.

The Evidence Boundary

Let’s be precise about what the evidence covers.

QINLOCK was studied in adults with advanced GIST previously treated with imatinib, sunitinib, and regorafenib. INVICTUS randomized 129 patients. The primary endpoint was progression-free survival, and the trial demonstrated a meaningful PFS advantage.

Dose escalation to 300 mg after progression has some support in clinical practice and literature. We built an exception pathway for it.

450 mg? That’s off the map. Not contraindicated—just unstudied. You’re extrapolating from extrapolation. The patient might benefit. Nobody knows, because nobody has run the trial.

The ethical point isn’t that off-label use is wrong. It’s that off-label use should move through a formal exception process with documented rationale—not an executive shortcut.

The Real Failure Mode: Override Culture

I want to name what happens, because it’s not fraud and it’s not cartoon villainy. It’s something painstakingly more boring: system capture.

Override culture is when exceptions get initiated through influence channels instead of clinical ones. When they’re granted without the same evidentiary burden. When they’re documented thinly—or not at all. When they’re insulated from audit because the people who would audit them are exposed to the business consequences.

It’s corrosive in ways that don’t show up on a dashboard.

Here’s what it looks like: leadership offloads discomfort onto process owners. A commercial leader frames it as “just take another look.” A medical director feels the weight of a potential account loss and searches for a way to document it as defensible. A pharmacist sees “expedite” and understands what it means. Nobody made a single corrupt decision. Everyone made a small accommodation. And the system moved.

The people who feel it most are closest to the work. They’re the ones who later must explain to the next member why their exception didn’t qualify.

A plan can survive controversial decisions. What it can’t survive is a decision system that changes based on who is asking.

Exceptions Must Be Governed, Not Granted

I’m not arguing against exceptions. Exceptions are necessary. Clinical coverage reality is messy, and rigid criteria will always have edge cases that deserve flexibility.

But there’s a difference between a governed exception and a granted one. A governed exception has structure, documentation, and accountability. A granted exception has a phone call and a wink.

Separation of powers. The direct decision-makers in a commercial relationship should not have any influence over medical necessity. Sales teams can flag urgency. An employer group can have a representative speak on their behalf.

A standardized exception dossier. Every non-standard request requires a structured packet: clinical rationale with literature anchor (or explicit acknowledgment that strong evidence is lacking), patient-specific details, alternatives tried, risk-benefit framing, and discontinuation criteria. “What would make us stop?” should be answered before you start.

A standing exception committee. For high-cost or high-uncertainty cases, convene a small group: medical director, clinical pharmacist, compliance/ethics, and optionally an external specialist. Decisions should be documented, signed, and time-limited—approved for 60–90 days with a review date.

Provenance and audit trail by default. Every exception generates a decision memo: policy version referenced, reason for deviation, evidence accepted, who approved, and expiration date. This isn’t bureaucracy. It’s protection—for the plan, the patient, and the people making the call.

The point isn’t to block exceptions. It’s to prevent unaccountable exceptions from becoming shadow policy.

How to Say No Without Being Inhumane

Saying no to a dying patient or their family member is brutal. Saying no when an executive is on the line is terrifying.

But there’s a version of “no” that is defensible, compassionate, and holds the line:

“We can review this under our exception pathway. Here’s what we’ll need: clinical rationale, supporting literature, and attestation from the treating oncologist. We can turn it around in 48 hours.”

“We can’t make coverage decisions outside our governance process based on commercial considerations—that would mean treating this member differently than others in the same situation.”

“If you believe our criteria are outdated, we can fast-track a policy review. But today’s decision follows today’s governance.”

The tone matters. You’re not hiding behind process to avoid accountability. You’re explaining why process is accountability.

And sometimes you say all of this, and the exception still gets approved over your head. That’s when dissent becomes its own form of integrity.

Compassion isn’t saying yes because you’re scared. It’s building something that holds when the pressure comes.

The Uncomfortable Truth

In any pooled insurance system, some groups have more leverage than others. Large employers pay more in premiums, generate more revenue, and carry more negotiating power. That’s the business model.

The question is whether that leverage extends into clinical decisions.

Without explicit governance, “business risk” becomes a covert clinical criterion. The sales team knows which accounts matter. Clinical leaders know which denials generate escalations. Nobody has to say it out loud. The pressure is ambient.

That’s ethically indefensible. It converts coverage into status-based privilege.

Frame it as conflict-of-interest risk, compliance risk, reputational risk, or long-term financial risk—precedent creates expectation, and expectation creates cost. Frame it however gets traction. But don’t pretend it isn’t happening.

For a concrete example of how these dynamics play out, see Strensiq:An End-to-End FWA Risk Scenario

If your system allows leverage to substitute for evidence, you haven’t built stewardship. You’ve built a marketplace for exceptions.

What Stewardship Protects

Stewardship protects patients who genuinely meet criteria for coverage. When exceptions are governed tightly, standard approvals move faster—less suspicion, less friction.

Stewardship protects clinicians who practice within evidence. They don’t have to wonder whether careful documentation is being outmaneuvered by louder channels.

Stewardship protects the plan from overcorrecting later. Every ungoverned exception today becomes a tighter restriction tomorrow.

And stewardship protects trust—the thing that makes pooled insurance workable at all. Members trust rules apply equally. Employers trust they’re not subsidizing special deals. Clinicians trust criteria mean something.

We need governance designed for the moment the CEO calls. Not because the answer must always be no—but because we need no to still mean something.

The goal isn’t rigidity. The goal is legitimacy.

The ethical test of a health plan isn’t whether it can approve. It’s whether it can stay fair when saying yes is easiest.