Amalia Rosu, Medical Oncologist and Head of the Oncology Department, Ponderas Academic Hospital, Bucharest, Romania. Participates in multidisciplinary tumor boards, presents this article on the challenges of making treatment decisions when pathology, imaging, molecular testing, and the clinical picture do not point in exactly the same direction. The article explores a complex metastatic cancer case in which immunohistochemistry suggested a breast origin, while the dominant lesion and overall clinical presentation pointed toward lung cancer. It highlights the limitations of individual biomarkers, the value of negative molecular testing, and the importance of integrating pathology, imaging, molecular data, PD-L1 expression, and patient-specific factors in clinical decision-making.
The article also examines the role of clinical judgment in navigating uncertainty, emphasizing that evidence-based oncology is not simply an algorithm but evidence-guided, patient-centered decision-making under uncertainty. It explores how oncologists can acknowledge uncertainty while still making a defensible treatment decision based on the totality of available evidence.
The Oncologist Between Evidence and Uncertainty
When pathology, imaging and molecular testing do not tell the same story
There is a particular moment in oncology that cannot be found in any guideline. It comes when the evidence is available, the diagnostic tests have been performed, the treatment options are clearly listed — and yet, the right decision for the patient in front of you is still not obvious. This is where oncology becomes more than the application of evidence.
It becomes judgment.
A recent case reminded me of this uncomfortable truth. A woman in her late forties presented with an extensive metastatic malignancy involving the thoracic compartment and multiple distant sites, including the brain, bone, lymph nodes and several visceral and soft-tissue locations.
The imaging strongly suggested a primary lung malignancy. But the pathology told a more complicated story.
Biopsies were obtained from the dominant pulmonary lesion and from a distant lymph node site. The morphology and immunohistochemical profile were essentially identical, supporting the interpretation of a single metastatic disease process. The immunohistochemical profile showed strong and diffuse expression of GATA3 and TRPS1, with focal low-level estrogen receptor expression. Markers commonly used to support pulmonary, gastrointestinal, Müllerian or melanocytic origin were not supportive.
The profile strongly suggested breast origin.
And yet there was no convincing primary breast lesion on breast imaging. The dominant lesion was pulmonary, with extensive locoregional thoracic involvement and a clinical presentation highly compatible with a primary lung cancer.
The question was no longer simply:
“What is the diagnosis?”
It became:
“Which piece of evidence should determine the treatment?”
When pathology and anatomy disagree
Immunohistochemistry is extraordinarily powerful, but it is not synonymous with organ-of-origin certainty.
TRPS1 has become an important marker in the diagnostic evaluation of breast carcinoma, particularly in difficult and triple-negative cases1. However, several studies have demonstrated that TRPS1 expression can occur in a subset of non-breast malignancies, including lung tumors2,3. GATA3, similarly, is valuable but not completely specific4,5.
This matters enormously in metastatic disease. A strong immunophenotype can be persuasive. But persuasion is not the same as proof. In this patient, the combination of strong GATA3 and TRPS1 expression raised a legitimate possibility of breast origin. Yet the overall clinical and radiological picture pointed in another direction.
Rather than allowing one test to dominate the entire diagnostic process, the case required integration of morphology, immunohistochemistry, imaging, clinical presentation and molecular data.
That is not uncertainty caused by inadequate medicine.
It is uncertainty created by the complexity of cancer itself.
The molecular test was negative – and that was also information
Comprehensive molecular testing did not identify an actionable driver alteration. No targetable EGFR, ALK, ROS1, RET, NTRK, BRAF V600E or KRAS G12C alteration was identified. At first glance, a negative molecular panel may appear to provide little information. In reality, it changed the therapeutic landscape.
Modern oncology increasingly divides metastatic lung cancer into biologically defined groups. When an actionable driver is identified, targeted therapy can fundamentally alter the treatment strategy. When no actionable driver is identified, the treatment pathway is different. The current ASCO living guideline specifically addresses first-line management of metastatic NSCLC without driver alterations and supports chemo-immunotherapy approaches according to histology, PD-L1 expression and clinical circumstances6.
In this case, molecular testing did not give us a treatment. It told us which treatments were not appropriate. That is an important distinction.
A negative biomarker is not an absence of information. It is information that narrows the decision space.
PD-L1: another piece of the puzzle
The tumor demonstrated low-level PD-L1 expression. This did not provide a dramatic predictive signal.
It provided context. For metastatic nonsquamous NSCLC without actionable driver alterations, pembrolizumab combined with platinum-based chemotherapy and pemetrexed remains a guideline-supported first-line strategy across low or negative PD-L1 categories, with the strength of recommendation depending on the specific biomarker and clinical setting6. The important point is that the treatment decision was not based on PD-L1 alone.
It emerged from the convergence of several pieces of evidence: clinical presentation + radiology + pathology + molecular testing + PD-L1 + patient-specific factors. This is what precision oncology increasingly looks like in real life. Not one perfect biomarker. But the integration of imperfect information.
The patient is not a clinical trial
The case also contained another layer of complexity. The patient had a significant underlying hematologic profile, with elevated inflammatory markers accompanying the malignancy.
These findings mattered – not because they changed the tumor’s molecular classification, but because they changed the patient’s risk profile. The decision to treat could not be separated from the risk of thrombosis, infection, treatment tolerance and overall clinical condition.
A clinical trial may tell us that a regimen improves survival in a defined population. It cannot tell us exactly how that benefit should be balanced against the individual risks of the patient sitting across from us.
This is one of the most difficult transitions in oncology: moving from population-level evidence to individual-level decisions.
Evidence does not eliminate uncertainty
Evidence-based medicine is sometimes misunderstood as an algorithm. It is not. Guidelines are essential. Randomized trials are essential. Molecular diagnostics are essential. Expert consensus is essential.
But none of them eliminates the responsibility of clinical judgment.
The oncologist still has to ask:
- Is the diagnosis sufficiently secure to justify treatment?
- Which primary site best explains the totality of the evidence?
- Does the molecular profile open or close therapeutic options?
- Is the patient physiologically able to receive the treatment?
- What risks are we accepting?
- What matters most to the patient?
- And perhaps most importantly: what are we still uncertain about?
The last question is often the most difficult.
Sometimes the most responsible decision is to acknowledge uncertainty
There is a tendency in medicine to associate uncertainty with weakness. I believe the opposite can be true. Recognizing uncertainty can be a sign of clinical maturity.
In a complex metastatic cancer, saying “we know” when the evidence is contradictory can be more dangerous than saying “this is what the totality of evidence currently supports, but important uncertainty remains.”
That distinction matters. Because uncertainty should not lead to therapeutic paralysis. It should lead to better integration of evidence. In this patient, the goal was not to find one test that would magically resolve every contradiction.
The goal was to reach the most defensible clinical decision based on the totality of available evidence – while keeping the diagnosis and treatment strategy open to revision as new information emerged.
That is the essence of modern oncology.
Between guidelines and the patient
The practice of oncology is often described as evidence-based medicine. Perhaps a more accurate description would be:
evidence-guided, patient-centered decision-making under uncertainty.
The evidence gives us boundaries. The guidelines give us pathways. Molecular testing gives us biological clues. Imaging gives us the map. Pathology gives us another layer of identity. But the oncologist has to integrate all of them. And sometimes, the most important skill is not knowing which test to order next.
It is knowing how much weight to give each piece of evidence. The future of oncology will undoubtedly bring better biomarkers, more sophisticated molecular classification and increasingly individualized treatment.
But uncertainty will remain. Cancer biology is dynamic. Tumors evolve. Biomarkers have limitations. Clinical trials cannot capture every patient. And no guideline can anticipate every clinical scenario.
The oncologist’s role is therefore not to eliminate uncertainty. It is to navigate it responsibly. Because ultimately, the hardest oncology decisions are not made when there is no evidence.
They are made when there is enough evidence to act – but not enough to make the decision easy.
And perhaps that is where the real art of oncology begins.
Key Takeways
- A marker is evidence, not a diagnosis – GATA3 and TRPS1 strongly suggest breast origin, but neither is absolutely specific.
- A negative molecular panel is not a null result – it narrows the decision space and redirects the therapeutic pathway.
- PD-L1 is context, not a verdict – it only becomes meaningful in combination with histology and molecular status.
- The patient’s risk profile is a separate axis from the tumor’s biology, and must be weighed independently.
- Clinical judgment is not a failure of evidence-based medicine – it is what completes it.
References
- Ai D, Yao J, Yang F, et al. TRPS1: a highly sensitive and specific marker for breast carcinoma, especially for triple-negative breast cancer. Mod Pathol. 2021;34(4):710-719. doi:10.1038/s41379-020-00692-8
- Georgescu AC, Georgescu TA, Duca-Barbu SA, Pop LG, Toader DO, Suciu N, Cretoiu D. A comprehensive review of TRPS1 as a diagnostic immunohistochemical marker for primary breast carcinoma: latest insights and diagnostic pitfalls. Cancers (Basel). 2024;16(21):3568. doi:10.3390/cancers16213568
- Boulogeorgou K, Topalidis C, Koletsa T, Karayannopoulou G, Kanitakis J. Expression of TRPS1 in metastatic tumors of the skin: an immunohistochemical study of 72 cases. Dermatopathology (Basel). 2024;11(4):293-302. doi:10.3390/dermatopathology11040031
- Hashmi AA, Brogi E, Wen HY. Trichorhinophalangeal syndrome type 1 (TRPS1) in breast pathology: diagnostic utility and pitfalls. Diagn Pathol. 2025;20:35. doi:10.1186/s13000-025-01623-4
- Rammal R, Goel K, Elishaev E, Soong TR, Jones MW, Zhao C, Clark BZ, Carter GJ, Yu J, Fine JL, Villatoro TM, Harinath L, Bhargava R. Utility of TRPS1 immunohistochemistry in confirming breast carcinoma: emphasis on staining in triple-negative breast cancers and gynecologic tumors. Am J Clin Pathol. 2023;160(4):425-434. doi:10.1093/ajcp/aqad066
- Reuss JE, Bazhenova L, Ismaila N, et al. Therapy for stage IV non-small cell lung cancer without driver alterations: ASCO living guideline, 2026.3.0. J Clin Oncol. Published online February 3, 2026. doi:10.1200/JCO-25-02825
- Lew M, et al. Young investigator challenge: the utility of GATA3 immunohistochemistry in the evaluation of metastatic breast carcinomas in malignant effusions. Cancer Cytopathol. 2015;123(10):576-581. doi:10.1002/cncy.21574
- Chen YA, et al. The diagnostic utility of trichorhinophalangeal syndrome type 1 immunohistochemistry for metastatic breast carcinoma in effusion cytology specimens. Cancer Cytopathol. 2023;131(4):226-233. doi:10.1002/cncy.22663
Other articles featuring Amalia Rosu on OncoDaily.