22. Neuro-Oncology CT

Primary brain tumors generate their CT signatures through a small number of tissue-level mechanisms whose interplay the physician must learn to read backward from the image. Attenuation on non-contrast CT is a near-linear function of electron density, and in brain tumors that density is modulated chiefly by cellularity, calcification, hemorrhage, necrosis, and lipid or cyst content. Contrast enhancement is, by contrast, a pure surrogate for blood–brain barrier (BBB) breakdown and microvascular leak: iodinated contrast is excluded from the normal CNS interstitium by tight junctions, so enhancement on CT reports exactly where neoangiogenic, fenestrated vessels have replaced or disrupted that barrier. These two physical readouts — intrinsic density and barrier integrity — are orthogonal, and disciplined interpretation keeps them separate.

In the 2021 WHO classification, diffuse gliomas are stratified by molecular markers (IDH mutation, 1p/19q codeletion, CDKN2A/B status), but the CT phenotype still tracks grade. Glioblastoma, IDH-wildtype (WHO grade 4) is the prototype: an irregularly thick, nodular ring of enhancement (BBB breakdown at the proliferating margin) surrounding a central hypoattenuating zone of necrosis (low density from liquefaction), with extensive surrounding vasogenic edema that is hypodense because plasma ultrafiltrate dilutes the white-matter electron density. The enhancing rind is typically $>3$ mm and frankly nodular, and corpus-callosum crossing (the 'butterfly' pattern) is characteristic. Lower-grade IDH-mutant astrocytomas are commonly non-enhancing or faintly enhancing hypodense expansile masses; oligodendrogliomas (IDH-mutant, 1p/19q-codeleted) classically calcify — coarse, nodular, often gyriform calcification reaches $90$–$1000$+ HU and is the single most useful CT clue, present in $\sim70$–$90\%$. Embryonal tumors such as medulloblastoma are hyperdense on non-contrast CT precisely because dense small-round-blue-cell packing raises electron density above normal cerebellum — the same hyperdensity logic that underlies primary CNS lymphoma.

The systematic search pattern proceeds: (1) localize the epicenter as intra-axial vs extra-axial (CSF cleft, dural tail, gray-matter buckling, bone reaction); (2) characterize intrinsic density against gray matter on a brain window ($\text{WW}\,80/\text{WL}\,40$) — hyperdense solid components argue hypercellularity; (3) interrogate calcification and hemorrhage on a wider window; (4) assess enhancement morphology (thin/regular vs thick/nodular ring); (5) map edema and mass effect with attention to herniation. The ranked differential is governed by Bayesian priors. In an adult over 50 with a solitary necrotic ring-enhancing mass and no known primary, GBM dominates the posterior. A hyperdense, homogeneously and avidly enhancing periventricular mass in an immunocompetent patient should raise primary CNS lymphoma, whose pretest probability rises steeply in the immunocompromised — where it instead enhances heterogeneously and competes with toxoplasmosis. Coarse calcification with a frontal-lobe cortical-based mass shifts probability toward oligodendroglioma. The dominant mimics are abscess (smooth thin-walled ring, but the discriminator on CT is limited — restricted diffusion on MRI is decisive), tumefactive demyelination (open-ring enhancement), and subacute infarct or contusion. Management hinges on tissue: maximal safe resection plus chemoradiation (the Stupp regimen) for GBM versus chemotherapy-led strategies for codeleted oligodendroglioma. Failure modes: beam-hardening and partial-volume artifact at the skull base mask infratentorial and temporal-lobe tumors (the posterior-fossa 'streak' is the classic technical trap); volume averaging fabricates pseudo-enhancement; and the cognitive trap of anchoring on 'edema' lets an isodense low-grade glioma be dismissed as ischemia. CT is a triage and calcification/hemorrhage tool — contrast-enhanced MRI remains the reference standard for characterization and is mandatory before commitment to a diagnosis.

Cerebral metastases are the most common intracranial neoplasm in adults, outnumbering primary brain tumors by roughly an order of magnitude, and their CT behavior follows directly from hematogenous seeding at the gray–white junction. Tumor emboli lodge where the caliber of penetrating arterioles abruptly narrows — at the corticomedullary junction and in arterial watershed/border-zone territories — which is why metastases characteristically cluster peripherally and at the gray–white interface rather than deep in white matter. Once established, a metastasis recruits a neovasculature that lacks a competent blood–brain barrier; consequently virtually all macroscopic metastases enhance, and the disproportionate vasogenic edema sign — a small enhancing nodule with edema far out of proportion to lesion size — is a hallmark, reflecting the highly permeable, leaky neovessels.

On non-contrast CT most metastases are iso- to hypodense, but several primaries produce hyperdense deposits whose density is diagnostically loaded. Hyperattenuation arises from hypercellularity, hemorrhage, or intrinsic mineralization: melanoma and choriocarcinoma, renal cell and thyroid carcinoma classically hemorrhage (the 'great hemorrhagic mets'), while mucinous adenocarcinoma (colon, ovary, some breast) and osteosarcoma may calcify or be intrinsically dense. A hyperdense cerebral nodule with surrounding edema in a patient with a known visceral malignancy is metastasis until proven otherwise. Enhancement is usually solid and nodular when small and ring-like when central necrosis develops; the ring of a necrotic metastasis can be indistinguishable from GBM or abscess on morphology alone, so lesion multiplicity and clinical context carry the discriminating weight.

The search pattern is deliberately exhaustive because management pivots on count and total volume: inspect the entire cortical ribbon and gray–white junction systematically by vascular territory, scrutinize the posterior fossa (cerebellum is a favored site and the most artifact-degraded region on CT), check for leptomeningeal nodularity and communicating hydrocephalus, and always obtain post-contrast imaging. The Bayesian logic is stark: in a patient with a known extracranial primary, the pretest probability of metastasis for a new enhancing lesion is very high, and multiplicity drives the posterior decisively toward metastatic disease and away from a solitary primary glioma or lymphoma. Conversely, a single enhancing lesion in a patient with no known cancer is more often a primary tumor than a metastasis, and indiscriminate labeling as 'mets' is a classic premature-closure error. The dominant mimics of multifocal enhancing disease are septic emboli/abscesses, neurocysticercosis, tuberculomas, multifocal demyelination, and subacute infarcts (which enhance and can be multiple). Management and prognosis flow from number and burden: stereotactic radiosurgery is favored for limited intracranial disease (the historical 'oligometastatic' threshold of $\le 3$–$4$ lesions has loosened with modern multi-isocenter platforms toward total-volume-based decisions), whole-brain radiotherapy and increasingly CNS-penetrant systemic agents and immunotherapy address higher burden, and resection is reserved for large symptomatic or diagnostically uncertain lesions. The decisive failure mode is that CT sensitivity for small ($<5$ mm) and posterior-fossa metastases is markedly inferior to contrast-enhanced MRI, which detects more and smaller lesions and frequently reclassifies a 'solitary' CT lesion as multiple — directly changing therapy. Technical traps include posterior-fossa beam-hardening obscuring cerebellar deposits and inadequate contrast timing or volume blunting small-nodule conspicuity; the cognitive trap is satisfaction of search — finding one metastasis and stopping.

Interpretation of the postoperative brain is governed by a single overriding principle: the blood–brain barrier breaks down as a consequence of surgery itself, so contrast enhancement loses much of its specificity for residual tumor within days of resection. The reactive, granulation-tissue enhancement that develops at the margins of a resection cavity is thin, linear, and follows the cavity contour; it characteristically begins within days and can persist for weeks to months, and it is the reason the field standardizes an early postoperative MRI within 24–48 (ideally <72) hours to establish a baseline before reactive enhancement contaminates the picture. CT cannot substitute for that MRI baseline for residual-tumor assessment, but CT is the workhorse for the acute postoperative complication question and for patients who cannot undergo MRI.

The immediate postoperative CT is read as a complication survey. Hemorrhage into or adjacent to the resection bed is the most time-critical finding: acute blood is hyperdense ($\sim50$–$80$ HU), and the physician must distinguish expected small-volume blood products and hemostatic agents lining the cavity from an expanding hematoma exerting mass effect. A pitfall is that some topical hemostatic materials and air–blood–fluid levels mimic or obscure hematoma; the actionable discriminator is interval expansion and mass effect, not absolute density. Pneumocephalus is universal after craniotomy; the critical distinction is benign post-craniotomy air from tension pneumocephalus, in which air under pressure compresses and separates the frontal lobes — the 'Mount Fuji' sign — and constitutes a surgical emergency. Ischemia from retraction injury or sacrifice of a perforator or bridging vein appears as evolving hypodensity in a vascular or peri-cavitary distribution and is a leading cause of unexpected postoperative deficit; on early CT it may be subtle, and restricted diffusion on MRI is far more sensitive. Additional expected findings include cerebral and extra-axial fluid collections, dependent layering of blood and proteinaceous fluid in the cavity, and dural/scalp reconstruction hardware that generates streak artifact.

The Bayesian framing of residual versus reactive enhancement leans heavily on timing and morphology. Nodular, mass-like enhancement present on the immediate (<72 h) study, before reactive change can develop, is presumptively residual tumor; thin linear rim enhancement appearing on a study obtained a week or more out is presumptively benign granulation tissue. This is why a delayed first scan is treacherous: enhancement obtained at, say, 5–30 days postoperatively cannot be reliably separated into tumor versus surgical reaction, and the baseline window is therefore narrow by design. Management consequences are immediate — an expanding hematoma or tension pneumocephalus prompts re-operation, whereas expected postsurgical change prompts observation and adherence to the imaging schedule. Failure modes are both technical and cognitive: metal and bone-flap beam-hardening artifact obscures peri-cavitary hemorrhage and residual tumor; volume averaging at the irregular cavity wall fabricates pseudo-nodularity; and the dominant cognitive error is mistaking expected postoperative enhancement for residual or recurrent tumor (or the reverse) when the scan is obtained outside the protocolized baseline window. The corollary teaching point is that the postoperative baseline, once correctly timed, becomes the anchor against which all subsequent surveillance — and every later progression call — is measured.

Radiation injury to the brain is a dose-, volume-, and time-dependent continuum whose imaging expression the physician must place on a temporal axis to interpret correctly. The unifying pathophysiology is vascular and glial: ionizing radiation injures endothelium and oligodendroglial precursors, producing a delayed obliterative vasculopathy with fibrinoid necrosis, hyalinization, and small-vessel thrombosis, together with demyelination and white-matter coagulative necrosis. Because these are slow biological processes, the cardinal clinical-imaging rule is temporal: injury is classified as acute (during/within weeks of therapy, usually transient edema), early-delayed (weeks to a few months, often transient demyelination and the pseudoprogression phenomenon addressed separately), and late (months to years), which is typically irreversible and includes radionecrosis.

Radiation necrosis is the defining late lesion. On CT it presents as a focal region of low attenuation with mass effect and, after contrast, irregular or 'soap-bubble'/'Swiss-cheese' enhancement that arises within or adjacent to the prior high-dose volume — the spatial constraint to the radiation field, reconstructed from the treatment plan, is among the most useful discriminators available. The central diagnostic problem is that radionecrosis and recurrent tumor are frequently indistinguishable on anatomic CT or MRI: both produce enhancing, edematous, mass-like lesions, because both reflect BBB breakdown. The physician therefore reasons probabilistically using time-since-radiation (a peak incidence of necrosis around $6$–$18$ months after fractionated radiotherapy, and earlier and more frequently after stereotactic radiosurgery), location relative to the dose distribution, and temporal behavior on serial imaging. Beyond focal necrosis, the spectrum includes diffuse leukoencephalopathy (confluent symmetric periventricular white-matter hypodensity with atrophy and ventricular enlargement, worse with concurrent chemotherapy and whole-brain radiotherapy), mineralizing microangiopathy (basal-ganglia and subcortical calcification, classically after combined therapy in children), radiation-induced cavernous malformations and telangiectasias (a delayed hemorrhagic risk best shown by MRI susceptibility imaging), large-vessel radiation vasculopathy with accelerated atherosclerosis and moyamoya-like stenoses, SMART syndrome (stroke-like migraine attacks after radiation therapy, with transient gyriform enhancement), and radiation-induced second neoplasms (meningiomas, sarcomas, gliomas) years to decades later.

The Bayesian differential for a new enhancing lesion in an irradiated brain is anchored by dose-volume context and latency. A lesion arising inside the $\ge 50$–$60$ Gy isodose region $9$–$15$ months after chemoradiation, with feathery 'spreading-wavefront' or soap-bubble enhancement and a comparatively low metabolic and perfusion signature, shifts the posterior toward necrosis; a lesion at the original tumor margin with rising enhancing volume, restricted diffusion, and elevated relative cerebral blood volume shifts it toward recurrence. Because anatomic imaging is insufficient, adjunctive physiology is decisive: MR perfusion (recurrent tumor shows higher relative cerebral blood volume than necrosis), MR spectroscopy (tumor elevates choline and the choline/NAA ratio, whereas necrosis tends toward a depressed, lipid-lactate profile), and amino-acid PET (e.g. FET/MET, increased uptake in tumor) materially move the probability. Management differs fundamentally — necrosis is treated with corticosteroids, bevacizumab (which reduces vasogenic edema by restoring vascular integrity), and occasionally surgery or laser interstitial thermal therapy, whereas recurrence triggers oncologic re-treatment — so the error cost of confusion is high. Failure modes: the principal technical limitation is that CT lacks the contrast resolution and the perfusion/diffusion/spectroscopic tools to separate necrosis from tumor, mandating MRI and often PET; calcification from mineralizing microangiopathy can be mistaken for tumoral calcification; and the cognitive trap is failing to register the latency and dose-field context, the two priors that most strongly inform the call.

Two treatment-induced phenomena dominate modern neuro-oncologic imaging because they decouple the contrast-enhancement signal from true tumor burden in opposite directions: pseudoprogression and pseudoresponse. Both are iatrogenic artifacts of effective therapy, and misreading either propagates directly into wrong treatment decisions, which is why the field abandoned the enhancement-only Macdonald criteria in favor of the RANO framework.

Pseudoprogression is a subacute treatment effect seen most characteristically after the Stupp regimen of concurrent temozolomide and radiotherapy for glioblastoma. Pathophysiologically it is transient radiation- and chemotherapy-potentiated injury to the tumor and adjacent brain — increased vascular permeability, inflammation, and incipient necrosis — that produces new or enlarging contrast enhancement and edema, typically within the first 12 weeks (and especially within the first 4–8 weeks) after completing chemoradiation, in the absence of true tumor growth, with subsequent stabilization or regression on continued therapy without a change in treatment. It is more frequent and more pronounced in tumors with MGMT promoter methylation, the very molecular subset that responds best — a striking coupling of good biology to a confounding image. Clinically and radiologically it is indistinguishable from early true progression on a single anatomic study; the operational solution embedded in RANO is the 12-week rule: progression should generally not be declared on enhancement appearing within 12 weeks of chemoradiation unless the new enhancement is largely outside the radiation field or there is unequivocal histopathologic proof. Perfusion (pseudoprogression tends toward lower relative cerebral blood volume), diffusion, spectroscopy, and amino-acid PET are deployed as adjuncts, and serial follow-up remains the arbiter.

Pseudoresponse is the mirror-image trap introduced by anti-angiogenic therapy, principally the VEGF inhibitor bevacizumab (and VEGFR tyrosine-kinase inhibitors). By restoring blood–brain barrier integrity, these agents reduce contrast leakage within hours to days, producing a rapid, dramatic decrease in enhancement and edema that overstates true anti-tumor effect — a 'normalization' of vasculature rather than tumor kill. The tumor may continue to grow as a non-enhancing, infiltrative T2/FLAIR-hyperintense component even as the enhancing volume shrinks, and this infiltrative progression carries poor prognosis. The teaching consequence is foundational: under anti-angiogenic therapy, contrast enhancement is an unreliable biomarker of tumor burden, and assessment must incorporate the non-enhancing disease and clinical status. The dominant mimics/confounders that masquerade as treatment effect or its resolution include corticosteroid-induced reduction in enhancement and edema (steroids must be accounted for when grading response), postoperative reactive enhancement, and radiation necrosis on the progression side. Management implications are direct and high-stakes: calling pseudoprogression 'progression' can lead to premature, erroneous abandonment of an effective regimen or unnecessary reoperation, whereas calling pseudoresponse a genuine 'response' can mask ongoing infiltrative growth and delay a needed change in therapy. Failure modes include the technical insensitivity of CT to the non-enhancing T2/FLAIR component that betrays pseudoresponse (an MRI-dependent observation), inconsistent contrast dose/timing across serial studies that fabricates apparent change, steroid-confounded enhancement, and the cognitive errors of anchoring on the enhancing volume and ignoring the latency window after chemoradiation.

Adjudicating true tumor progression is the synthesis toward which the preceding sections build: it requires separating genuine neoplastic growth from postoperative change, radiation necrosis, pseudoprogression, pseudoresponse, and steroid effect, using a standardized framework rather than an impressionistic read of 'more enhancement.' The governing standard is the Response Assessment in Neuro-Oncology (RANO) criteria — and its immunotherapy adaptation iRANO — which replaced the older Macdonald system precisely because enhancement-only assessment proved unreliable in the era of chemoradiation and anti-angiogenic and immune therapy.

RANO grades response across four categories (complete response, partial response, stable disease, progressive disease) using bidimensional measurement of enhancing disease, but its conceptual advance is twofold: it formally incorporates the non-enhancing T2/FLAIR component as a marker of infiltrative tumor (critical for detecting pseudoresponse), and it incorporates corticosteroid dose and clinical status into the response determination. Progressive disease under RANO generally requires a $\ge 25\%$ increase in the sum of the products of perpendicular diameters of enhancing lesions (on stable or increasing steroids), a significant increase in non-enhancing T2/FLAIR disease not attributable to other causes, any new lesion, or unequivocal clinical deterioration. Embedded in the framework are the temporal safeguards already discussed — the 12-week post-chemoradiation window within which new enhancement inside the radiation field is presumed pseudoprogression unless proven otherwise — and, for checkpoint-inhibitor therapy, iRANO extends this protective logic: apparent radiographic progression within the first 6 months of immunotherapy, in a clinically stable patient, should be confirmed by a follow-up scan approximately 3 months later before progression is declared, because immune-mediated inflammatory pseudoprogression can transiently enlarge or create enhancing lesions.

The Bayesian and physiologic reasoning that operationalizes progression integrates priors and adjuncts. Growth of enhancement at the original tumor margin (rather than diffusely in the dose field), accompanied by restricted diffusion (hypercellularity), elevated relative cerebral blood volume on perfusion, elevated choline/NAA on spectroscopy, and increased amino-acid PET uptake, collectively shift the posterior strongly toward true progression; the opposite profile — feathery enhancement confined to the high-dose region within the latency window, low rCBV, lipid-lactate spectroscopy — shifts it toward treatment effect. No single sign is pathognomonic, and the disciplined reader weights them against the timeline and the therapy. Management stakes are explicit and bidirectional: a false-positive progression call triggers premature abandonment of effective therapy, unnecessary reoperation, or inappropriate enrollment in salvage trials, whereas a false-negative delays salvage surgery, re-irradiation, or a change in systemic therapy and forfeits the window in which intervention helps. The failure modes are the cumulative set this chapter has built toward: technically, the insensitivity of CT to the non-enhancing T2/FLAIR component and to diffusion/perfusion/spectroscopic physiology makes CT inadequate as the primary tool for progression — contrast-enhanced multiparametric MRI is the reference standard, with PET as adjunct — while inconsistent contrast dose and timing, slice-misregistration, and beam-hardening corrupt serial comparison; cognitively, anchoring on enhancement, satisfaction of search, steroid-confounded measurements, and disregard of the post-treatment latency windows are the recurring errors. The mature interpretive posture treats every progression determination as a probabilistic judgment, made against a correctly-timed baseline, using a standardized framework, and explicitly hedged where treatment effect remains in the differential.