Every week, in roughly seventy academic cancer centers, a group of people sits in a room and decides how cancer should be treated.
A medical oncologist, a radiation oncologist, a surgeon, a pathologist, a radiologist, sometimes a molecular biologist, a genetic counselor, a pharmacist, a nurse navigator. They put a case on the screen. They argue. Somebody says the thing that changes the plan.
That room is where the standard of care is actually made, as opposed to where it is written down.
Roughly 80 percent of American cancer patients are not treated by anyone in that room. They are treated by community oncologists, in practices that do not have a molecular pathologist to hand, and frequently do not have a tumor board for the rarer diagnoses at all.
An analysis in JMIR Cancer described the situation precisely: expert tumor board insights "exist in silos and do not reach the wider oncology community."
So a community oncologist with an ambiguous case, an unusual molecular profile, or a rare tumor has three options.
Refer the patient to an academic center, and frequently lose them permanently. Phone a friend from fellowship, if she has one at the right institution. Or decide alone.
The knowledge she needs is being discussed this Thursday, in a room she is not in, by people she may well have trained with.
Community oncologists have said what they want, repeatedly
The most useful evidence here comes from asking community oncologists directly, and the answers are consistent and specific.
From a Cancer Medicine study surveying 85 oncologists with an 87 percent response rate:
- 74 percent prefer to learn about clinical trials through personal communication with investigators. Not databases. Not portals. A person.
- 82 percent report only minimal to moderate understanding of trial availability in their own region.
- They cite "delay for consults and subsequent lack of return communication" as a reason they hesitate to refer.
- And revealingly, 52 percent of tertiary center respondents described community practices primarily as a source of referrals for clinical trials.
Read those last two together. The community oncologist experiences the academic center as a place that takes patients and does not communicate back. The academic center describes the community practice as a referral source.
Neither is describing a peer relationship, and that is the whole problem.
What community oncologists say they want is not transfer of the patient. It is access to the expertise while keeping the patient. Those are different requests, and the system offers only the first.
Virtual tumor boards work, which makes the absence harder to explain
If this were a hard problem technically, the persistence would be understandable. It is not.
From a study of a virtual head and neck multidisciplinary tumor board published in OTO Open:
- More than 96 percent of participants found it useful.
- 84 percent said it improved access to other specialties.
- 64 percent reported shorter time to adjuvant care.
Other reported experience is similar. Attendance at one institution's board reportedly rose 46 percent after moving virtual. Participation becomes easier, not harder, when geography is removed.
And these are not marginal effects. Shorter time to adjuvant therapy is a clinical outcome.
So virtual multidisciplinary review is feasible, well received, and associated with better process measures. It exists in pockets, usually where a specific person championed it and a specific grant funded it. When the champion moves or the grant ends, it stops.
The constraint is not technology and not evidence. It is the invitation.
Why the door stays closed
Tumor boards are institutional property, and inviting outsiders raises real questions that nobody has an incentive to resolve.
Liability. If a community oncologist presents a case and the board offers an opinion, what is the board's relationship to that patient? Who documented what? The institution's counsel has a clear preference and it is not "invite strangers."
Credentialing. Participants in an institutional conference are typically credentialed at that institution. An outside physician is not, and the mechanism to include them cleanly does not exist in most medical staff structures.
Competition. An academic center inviting community oncologists into its tumor board is helping them keep patients that might otherwise be referred in. The institutional interest points the other way, even where individual physicians would happily help.
And documentation. Institutional boards generate a record that becomes part of care. An advisory discussion about an outside patient does not fit that structure.
Every one of these is solvable with clear design. None of them will be solved by an institution acting alone, because the benefit accrues to a physician somewhere else.
So access is granted personally rather than structurally. The community oncologist who trained at that center, or whose fellowship classmate is now on the faculty, gets her case discussed as a favor. The one who trained elsewhere does not.
The route in is whoever you know, and that graph is invisible to everyone including the people who could open the door.
What partial solutions reveal
Several things have grown up in the gap, and each one shows what the gap actually is.
Curated question and answer platforms. One oncology-specific platform reported 7,381 oncologists across 3,515 institutions, 5,131 of them in community settings, engaging with curated tumor board questions. That community-to-academic ratio tells you where the demand is. Its limitation is that it is asynchronous and curated, which serves the common question well and the specific hard case less well.
Paid virtual second opinions from major cancer centers. These work and they are a consumer product, purchased for an individual patient, rather than a professional consultation between clinicians.
Grant-funded regional virtual boards. Several have been built and evaluated positively. They depend on a champion and typically end with the funding.
Guidelines. NCCN and equivalents are excellent and are, by construction, general. The cases that need a board are the ones the guideline does not settle.
Each of these serves part of the need. None of them does the thing the community oncologist actually asked for, which is a named peer who will look at this specific case, this week, and who she can go back to.
Why precision oncology makes this worse
The timing matters, because the trend is running against the community physician.
Cancer treatment is becoming molecularly specific. Decisions increasingly turn on a particular variant, a particular expression pattern, a particular combination that a given oncologist may see a handful of times in a career.
Precision oncology has made expertise more tumor-specific and therefore rarer, which means the number of cases where a community oncologist genuinely needs a sub-specialist opinion is rising, not falling.
At the same time, the evidence on who reads a case keeps strengthening. Specialist second reads of head and neck cancer imaging changed the stage in 56 percent of cases and management in 38 percent. Who looks at the case matters enormously, and the community oncologist's access to that second look is determined by her personal network.
The molecular tumor board problem, in miniature
If you want to see the whole failure compressed into one example, look at what has happened with molecular tumor boards over the last five years.
Genomic sequencing of tumors became routine faster than the expertise to interpret it became distributed. A community practice can now order comprehensive genomic profiling on any patient, and frequently does. The report comes back with a list of variants, some actionable, some of uncertain significance, and a set of trial matches that may or may not be geographically plausible.
Interpreting that report well requires a specific and genuinely rare combination: molecular pathology, current knowledge of which agents have data in which contexts, and awareness of what is actually enrolling nearby. Academic centers built molecular tumor boards precisely because no individual oncologist can hold all of that.
The test is now universally available. The board that makes the test useful is not.
This is the credential-expertise gap in its sharpest form. Every oncologist ordering the test is board-certified in oncology. Almost none of them has the specific interpretive expertise the test now demands, because that expertise did not exist when they trained and lives in a small number of people.
And it will get worse rather than better, because the pace of new molecular findings exceeds the pace at which any individual can absorb them. The gap between what the technology produces and what the average practitioner can interpret is widening structurally.
Which means the value of routing a question to the person who does hold that expertise is rising every year, in a system where the only route is knowing them personally.
What would work
Route the question, not the patient. This is the core reframe and it is what community oncologists have been asking for. A structured, de-identified case presented for advisory review, with the patient staying in the community for treatment. Reported experience with virtual boards suggests exactly this effect: improved patient retention for non-surgical care.
Match on tumor sub-specialty, not institution. The right reviewer for a rare sarcoma variant is the person who sees them, wherever they work, which is frequently not the nearest academic center.
Use the fellowship graph as the opening. The community oncologist's fellowship classmates are now scattered across NCI-designated centers. That is a real, verifiable, high-trust connection to exactly the rooms she cannot enter, and nothing currently surfaces it.
Get the guardrails right, explicitly. De-identified case framing. A documented advisory rather than treating role, with clinical responsibility remaining unambiguously with the community oncologist. Patient consent where identifiable material is needed. These are the questions institutional counsel asks, and they have answers, but the answers have to be written down before anyone will participate.
Close the loop back. The stated reason community oncologists hesitate to refer is lack of return communication. Any system that does not guarantee the answer comes back reproduces the exact failure that made them stop referring.
What you can do now
If you are a community oncologist
Map your fellowship class. Where is everyone now? A meaningful number are at academic centers, on tumor boards, and would take your call. Most physicians have never assembled this deliberately.
Ask to attend, specifically. Many academic tumor boards will admit an outside physician who asks directly, particularly for a case they are presenting. The barrier is frequently that nobody asks, because nobody knows they can.
Ask for the sub-specialist by name. "Who at your center handles this variant" is a better question than "can I refer this patient," and it frequently produces a phone call rather than a transfer.
Say what you want up front. Community oncologists in the research consistently want expertise while retaining the patient. Saying that explicitly changes the conversation, because the academic side frequently assumes a transfer is being requested.
If you are at an academic center
Recognize what the survey data says about you. Fifty-two percent of tertiary respondents described community practices as a trial referral source. Your community colleagues know this, and it is why 82 percent have poor awareness of your trials and hesitate to send patients.
Invite people in. Your tumor board can accommodate an external presenter, and the institutional obstacles are real but not insurmountable if someone decides to solve them once rather than case by case.
Close the loop, always. The single most cited reason community oncologists hesitate to refer is that nothing comes back. That is entirely within your control and costs almost nothing.
If you lead a cancer program or a society
Fund the virtual board permanently, not as a grant. The evidence is favorable and the failure mode is entirely predictable: they end when the funding does.
Publish the guardrails. A model framework covering liability, documentation, consent, and advisory role for external participation in tumor boards would unlock a substantial number of institutions that want to do this and have no template.
Frequently asked questions
What percentage of cancer patients are treated in community settings? Roughly 80 percent of US cancer patients are treated by community oncologists rather than at academic centers, while the multidisciplinary tumor boards that shape standards of care are concentrated at approximately 70 NCI-designated cancer centers.
Do virtual tumor boards work? The published experience is positive. A study of a virtual head and neck multidisciplinary tumor board found more than 96 percent of participants rated it useful, 84 percent said it improved access to other specialties, and 64 percent reported shorter time to adjuvant care. Reported attendance also increased when boards moved to virtual formats.
Why can't community oncologists join academic tumor boards? Access is granted personally rather than structurally. Institutional concerns about liability, credentialing of non-affiliated participants, documentation of advisory opinions, and competition for patients mean no general mechanism exists, so participation depends on knowing someone at the institution.
What do community oncologists say they want? Expertise without losing the patient. Survey research found 74 percent prefer to learn about trials through personal communication with investigators, 82 percent report only minimal to moderate awareness of regional trial availability, and lack of return communication after referral is a cited reason for hesitancy to refer.
Does it matter who reviews a cancer case? Substantially. Specialist neuroradiologist second reads of head and neck cancer imaging changed stage in 56 percent of cases and changed management in 38 percent, indicating that the identity and sub-specialization of the reviewer materially affects the treatment plan.
How does precision oncology change this? It makes the problem larger. As treatment decisions turn increasingly on specific molecular findings, the relevant expertise becomes narrower and rarer, so the number of cases where a community oncologist genuinely needs a sub-specialist opinion rises while access remains dependent on personal networks.
The bottom line
The room where cancer treatment decisions are refined meets weekly at roughly seventy institutions. Eight in ten American cancer patients are treated by physicians who are not in it.
The technology to include them has existed for years, works well, is rated useful by more than 96 percent of participants, and is associated with faster adjuvant care. It exists in scattered pockets, dependent on individual champions and time-limited grants.
Community oncologists have said clearly what they want: the expertise, not the transfer, with the answer coming back. What they are offered instead is a referral pathway that frequently takes the patient and returns no communication, which is why they stop using it.
And the difference between a community oncologist whose hard cases get expert review and one whose do not is, overwhelmingly, whether she happened to train at the right place or knows someone who did.
The door is not locked. There is simply nobody whose job it is to open it, and no way for the person outside to find out who could.
Part of a series on the missing professional infrastructure of healthcare. Previously: Successor-less Closure
Evidence note: sources include JMIR Cancer (2022) on the siloing of tumor board expertise and community oncologist platform usage; Cancer Medicine (2021) surveying 85 oncologists on trial awareness and referral behavior; OTO Open (2023) on virtual multidisciplinary tumor board evaluation; the Journal of Otolaryngology (2013) on neuroradiology second reads; and secondary reporting on virtual tumor board attendance. The estimate that community oncologists treat 80 percent of cancer patients is widely cited in the oncology literature and definitions of community versus academic settings vary by source.
