Pulmonary Hypertension: methods & interpretation limits

Public module and private inputs

Pulmonary Hypertension is public, including its teaching chapters, RHC lab, risk tools, structured assistant, Current Evidence figure and evidence inventory. No section password or PH session is required.

Ask the exPHert AI chat retains a separate AI password and signed session shared with the Lab chat. The teaching cases retain their learning-access password; opening this module does not unlock cases.

Calculator and structured-assistant entries remain in browser page memory and clear on reload. Public module access does not add input storage or transmission. Deliberately submitting an AI question sends that question and bounded conversation history through the server to OpenAI, as described in the AI chat article.

Choose a workspace and keep your comparison

Learn, CTD & scleroderma, RHC lab, Ask the exPHert, Risk and Evidence answer different questions. Resources opens Drug mechanisms, Visual library and Sources & methods. Chapter search filters the learning library; evidence search and filters operate on study reports. Detailed appraisals and All retrieved records distinguish the editorial trial library from the wider screening inventory.

Visited RHC, risk and structured-review workspaces remain mounted when switching sections, preserving entries and comparisons in this page session. Reload clears those entries. Shared evidence loads when a workspace needs it; Try again retries an unsuccessful load rather than supplying incomplete evidence as a finished result.

Connective tissue disease and systemic sclerosis tools

CTD & scleroderma separates detection of possible PAH, systemic sclerosis classification, established SSc-ILD extent and prognosis after confirmed PAH. ASIG and DETECT are SSc screening approaches, not mortality scores. They do not establish PAH or justify pulmonary vasodilators. CTDs without established SSc require individualized assessment rather than automatic application of these SSc algorithms; established SSc may coexist with another CTD. Annual PAH assessment is recommended in SSc; unexplained breathlessness after noninvasive investigation can warrant RHC despite a reassuring screen.

ASIG is positive when NT-proBNP is ≥210 pg/mL (equivalent to ng/L), or haemoglobin-corrected DLCO is <70% predicted together with FVC%/DLCO% ≥1.8. BNP cannot be substituted. A positive arm can trigger evaluation while the other remains missing; a negative result requires complete valid inputs and confirmation of corrected DLCO. FVC/DLCO uses percent-predicted values from the same assessment, not litres divided by an absolute transfer factor. No sensitivity, prevalence or individual disease probability is generated.

The DETECT guide checks the guideline population: adult SSc, >3 years since the first non-Raynaud symptom, FVC ≥40% and DLCO <60% predicted. Original study exclusions and clinical context still matter. It does not recreate raw-variable risk points. Enter totals obtained from the published nomogram: Step 1 >300 prompts echo; only then is Step 2 interpreted, with >35 prompting RHC. Exactly 300 and 35 do not cross those published thresholds. No missing-variable imputation or alternative >40 cutoff is implemented. The discontinued official website is not offered as a working calculator.

The 2013 ACR/EULAR classification tool uses category maxima and a ≥9 threshold. Bilateral finger thickening proximal to MCP joints is sufficient (9 points); other items are not added to that sufficient criterion. Finger-sparing skin thickening or a better scleroderma-like explanation makes the criteria inapplicable. Unknown applicability blocks a classification result. An observed subtotal and possible total retain missing findings; a lower classification score does not exclude early SSc. PAH and/or ILD earns only one 2-point category; an echo pressure estimate alone is not established PAH.

Goh extent staging requires established SSc-ILD. Clearly limited (<20%) or extensive (>20%) HRCT extent is used directly. If expert HRCT assessment is indeterminate (approximately 10–30%), FVC <70% indicates extensive and FVC ≥70% limited disease. A raw CT percentage is not automatically routed through contradictory overlapping ranges. This is ILD prognostic context, not a PAH mortality estimate or a rule assigning group 1 versus group 3. Current ERS/EULAR CTD-ILD guidance supports HRCT screening in all SSc; PFTs do not replace HRCT.

After RHC and clinical phenotyping establish group 1 PAH, use the separate three-strata, four-strata/COMPERA and REVEAL workspaces. Workup must consider ILD, left-heart disease, thromboembolism and venous/capillary involvement. The tools remain a software prototype requiring clinical validation. Entries stay in page memory, survive section switching, and clear on reload; they are not submitted to AI.

CTD literature: study design, population and uncertainty

CTD & scleroderma → Literature review is a focused narrative reading guide to screening, prognosis, confirmed PAH treatment, early pulmonary vascular disease, ILD-associated PH and guideline recommendations. Each report keeps its population, design, endpoint/result, clinical interpretation, limitations and primary-source link. Topic selection, text search and finding labels support navigation. The displayed review date describes this selection; it is not a claim of an exhaustive systematic search or a pooled meta-analysis.

Direct SSc trials, CTD or SSc analyses of randomized parent trials, general PAH results that included CTD, observational cohorts and guidelines are explicitly distinguished. The finding label describes the stated endpoint in the stated report. A positive composite or exercise endpoint is not relabeled as isolated survival benefit; a subgroup is not a separate randomized trial. Registry survival is not an individual risk probability, and related reports can share participants.

Negative and neutral findings remain visible alongside supportive evidence. Rituximab and EDITA retain their neutral primary comparisons despite exploratory signals. EDITA-ON remains a nonrandomized follow-up rather than a second prevention trial. SPHInX retains severe under-recruitment and wide uncertainty. Scleroderma Lung Study II did not demonstrate between-treatment superiority; that result is neither proof of equivalence nor evidence for PAH immunotherapy. ILD trials and group 3 PH-ILD studies do not automatically support treating group 1 PAH or vice versa.

This reading guide is separate from the calculator logic and conditional medication rules. Adding or reviewing a report does not change calculator thresholds, establish a diagnosis or create an automatic treatment recommendation. The wider PH Evidence library remains available for full appraisals and related reports. Literature loads when its view is first opened, retaining the existing browser-only calculator workflow.

Visual explanations and retained guideline content

Original schematics explain circulation, diagnostic grouping, investigation flow, RV adaptation, treatment pathways and evidence. They simplify relationships for teaching; their geometry, colors, arrows and illustrations are not clinical images or measured anatomy. Each Learn lesson pairs its visual with a visible explanation, key teaching bullets and the clinical connection. Additional supporting points can be expanded in place; the page-wide reading control reveals the complete retained text. Citations remain beside the relevant lesson. Text and visuals stack on smaller screens, so the core explanation does not depend on a mouse hover.

The trial library is an extensive curated index across PAH, left-heart disease, lung disease, CTEPH and selected group 5 conditions, with its literature-search date displayed in Evidence. It includes formative and pivotal randomized studies, pediatric reports, investigational failures, informative extensions, single-arm studies and selected observational cohorts. Explicit editorial indexes filter trials by the tested drug or class, enrolled disease state and prespecified primary-outcome result. Background-only treatments remain separately labeled and do not become independently tested interventions. A second appraisal axis retains harm, exploratory evidence and mixed benefit-risk even when a primary surrogate endpoint was met. An endpoint, effect estimate, limitation and practical interpretation accompany each summary. Clinical group membership is explicitly assigned and may include several groups for mixed populations. Population-scope labels distinguish RHC-confirmed PH cohorts from adjacent or noninvasively selected cohorts where PH was not uniformly confirmed. The index searches verified DOI, PubMed, ClinicalTrials.gov and EU CT registration identifiers alongside names and clinical terms. Context filters separate controlled reports from extensions, single-arm and observational studies, registry results, conference abstracts and sponsor reports. Twenty reports render initially, with progressive disclosure for longer result sets. Counts describe reports, not independent cohorts or pooled participants. This selection is not a systematic review of every published or ongoing trial.

Independent real-world cohorts have their own observational context and a quick filter; an open-label interventional extension is not automatically a real-world cohort. OPUS/OrPHeUS, SPHERE, EXPOSURE and EXPERT supply routine-care exposure and safety context with explicit PAH subsets and overlapping populations. The pooled SPHERE/EXPOSURE risk analysis links to both registry reports without adding independent participants. Observational mortality associations, propensity adjustment and treatment persistence do not establish causal survival benefit. PH-ILD claims cohorts without uniform RHC confirmation remain distinct from invasively confirmed clinical cohorts; neither is an INCREASE extension. Later PATENT-2, pediatric ambrisentan, IMPRES imaging, ARIES-3 and EDITA-ON retain their parent results, selection and safety limitations.

Trial extensions, substudies & overlapping reports

The PH evidence index includes extensions, long-term follow-up, nested substudies, subgroup reports, secondary/post hoc analyses and pooled analyses. Report type is a separate editorial field from publication context, enrolled phenotype, overall appraisal and primary-outcome result. A subgroup of a randomized trial is not another independently randomized trial. An exploratory finding does not inherit the parent trial’s positive-primary label or overturn a failed primary result.

Each verified relationship has explicit parent report IDs and a cohort note. Open Trial family within a report to move to its related appraisals. The links traverse those explicit relationships, including reports with several lead-in studies; they are navigational groups, not proof that every participant overlaps across every linked report. Counts remain report counts. Do not sum their denominators, pool their effect estimates or interpret extension survival as an uncontrolled treatment effect.

Eligibility, randomized/enrolled/analysed/extension denominators, follow-up cutoff, maintained comparator, crossover, missing observations, multiplicity and selection bias are retained in the detailed cards. Corrections remain cited with the affected report and are separate candidate records in the source inventory; a shared DOI must not make a correction appear to be an independent result or silently give it the original appraisal. Trial-family coverage is verified incrementally and is not a claim that every worldwide companion publication has been appraised. The weekly watch explicitly chases extension and substudy reports without a prominence or outcome filter.

HFpEF mineralocorticoid antagonist evidence in Group 2

Finerenone, spironolactone, eplerenone and other MRA reports are indexed under Group 2 and their stated heart-failure phenotypes as evidence for the underlying left-heart condition. FINEARTS-HF includes mildly reduced as well as preserved ejection fractions; TOPCAT and Aldo-DHF used different LVEF thresholds. These studies did not uniformly require RHC-confirmed PH, so they retain the adjacent-population label and are excluded by the RHC-confirmed-only filter. They do not establish PH-specific efficacy or identical effects in isolated versus combined post-capillary PH.

Other MRA evidence retains its own design and eligibility: small eplerenone HFpEF trials, short balcinrenone safety or mixed-EF cardiorenal trials and uncontrolled esaxerenone cohorts do not receive finerenone’s clinical-outcome evidence by class association. Active comparators and prior-drug switching are distinguished from concurrently administered combinations. Shared dapagliflozin background is not indexed as a separately randomized intervention. The relevant HFmrEF/HFpEF stratum in a class-pooled analysis is distinguished from its larger all-EF population, and its parent trials are linked without counting their participants again. Unverified or off-phenotype studies remain in the source-review queue rather than acquiring invented HFpEF or PH outcomes.

The main composite, component outcomes, co-primary endpoints and safety findings are kept distinct. A positive recurrent-event composite is not proof of cardiovascular mortality benefit alone; a favorable secondary hospitalization finding does not make a neutral primary composite positive. Parent, subgroup, adherence, imaging, biomarker and other ancillary reports remain linked through Trial family with explicit overlapping cohorts and exploratory limitations. The evidence cards do not change simulator MRA responses, dosing, prescribing logic or risk algorithms.

Illustrated drug mechanisms

Seven original pathway diagrams explain endothelin receptor antagonism, PDE5 inhibition, soluble guanylate cyclase stimulation, IV/SC prostacyclin, oral prostacyclin-pathway therapy, sotatercept and calcium channel blockade. Arrows and inhibition bars describe simplified signaling relationships rather than measured effect sizes, dosing or a prediction of response. Parallel text explains each circuit, practical distinctions and source limitations.

Oral treprostinil is distinguished from the selective IP agonist selexipag; PDE5 inhibition and sGC stimulation are different actions on the cGMP system and are not a combined regimen. Sotatercept is represented as a ligand trap rather than a direct BMPR2 agonist or acute rescue vasodilator. IV epoprostenol and IV/SC treprostinil retain their separate routes. The clinical suitability of a mechanism depends on phenotype and evidence, not on the diagram alone.

RHC & therapies: preserved-baseline comparisons

The interactive pressure lab reuses the full simulator through mixedBaseline, supportedState and applyInterventions. rhythmPressure supplies the visible pressure contours; the display adds no physiological coefficients or separate drug-response curve. It holds respiration at the shared reference endpoint and keeps the native preset rate in sinus rhythm. Selecting a medication response always starts from the original state, rather than compounding changes to an already treated state. Reset restores the same baseline; changing the haemodynamic pattern resets the medication controls.

The catheter-location controls select RA, RV, PA or PAWP. Before and after contours share one fixed 0–100 mmHg axis, with the untreated reference dashed and the selected established response solid. Thin mean lines and the haemodynamic readouts use 2400 midpoint samples of one cardiac cycle. Optional ECG II is synchronized to the same rate. Hover, tap and keyboard inspection use the common waveform interpretation helpers. The equation mPAP = PAWP + PVR × CO describes the same coupled state; an adjacent pressure-budget graphic splits mean PA pressure into wedge and the PVR × CO gradient on a common scale. Displayed values are rounded.

The original catheter schematic follows a central venous route through SVC, RA, the tricuspid valve, RV, the pulmonary valve and PA. The wedge stage keeps the tip in a distal pulmonary artery branch: balloon occlusion permits a static pressure column toward the pulmonary veins and left atrium when measurement conditions are valid. The catheter does not enter the left atrium. The drawing is a simplified location guide, not measured anatomy, procedural instruction, a catheter-motion solver or a validation of wedge quality. PAWP is not automatically interchangeable with LV end-diastolic pressure.

ERA, PDE5 inhibition, IV epoprostenol, selexipag, oral treprostinil, SC treprostinil and sotatercept controls select bounded established-response snapshots from the existing pulmonary-therapy model. ERA, PDE5 inhibition and sotatercept use Off/On controls; On selects the existing representative established-response endpoint. Other therapies retain Low, Medium and High. Presets compare pathways rather than prescribe a sequence; none of these settings maps to clinical doses. Oral triple plus sotatercept selects an ERA, PDE5 inhibitor, oral treprostinil and sotatercept. The SC version selects an ERA, PDE5 inhibitor, continuous SC treprostinil and sotatercept. These are four-drug teaching comparisons, not validated regimen-effect predictions or a head-to-head trial of routes. Selexipag remains a separate selective IP agonist rather than being renamed an oral prostacyclin analogue. The RHC interface selects one IP-family agent at a time; externally supplied overlapping states remain bounded in the common model. Oral therapy represents established treatment after initiation and titration, the infusion represents a settled titrated state, and sotatercept represents later sustained follow-up around 24 weeks. No animation or control change models clinical onset, single-dose response, toxicity, safe infusion interruption or a validated patient forecast.

Pre-capillary, isolated post-capillary and combined profiles support comparison of resistance and filling-pressure mechanisms. The isolated post-capillary example receives no chronic PAH-drug PVR benefit in this model. In the combined example, falling PVR can coexist with rising wedge as pulmonary return loads a limited left heart. These synthetic experiments do not establish the clinical group, demonstrate group 2 treatment benefit or authorize PAH therapy. The separate medication model article documents the existing coefficients, shared-IP overlap and compliance approximation.

Right-heart catheter trace atlas

The atlas samples the existing circulation engine and its presets: reference [0], pre-capillary PH [1], severe mitral stenosis [8] for an isolated post-capillary example, and mixed pre-capillary PH plus dilated cardiomyopathy [1, 2] for a combined example. Each keeps its native preset heart rate and uses sinus rhythm. The sequences show RA, RV, PA and PAWP with two cardiac cycles at each location; all pressures share a fixed 0–100 mmHg axis. The atlas is synthetic educational output, not a patient tracing, a catheter-motion model or a new diagnostic simulation.

Pressure contours use rhythmPressure with respiratory time fixed at the reference endpoint and the scenario set to respiratory hold. No respiratory oscillation, intervention or catheter artifact is added in this atlas. Displayed cycle means use 2400 midpoint samples over one nominal cardiac cycle. PVR is calculated from those sampled mean PA and wedge pressures divided by the preset cardiac output, not from systolic PA pressure or the preset pressure labels. Cardiac output and native heart rate come from mixedBaseline. These numerical and display choices support consistent comparison; they are not newly validated clinical measurement methods.

The mitral-stenosis example illustrates why wedge is not interchangeable with LV end-diastolic pressure. A pre-capillary pattern does not identify its cause, and the combined example does not establish group 1 PAH or a treatment indication. The threshold labels follow the cited PH guideline; synthetic waveform amplitudes and model coefficients remain illustrative.

Risk methods and missing measurements

Risk assessment is intended for adults with an established group 1 PAH diagnosis, with clinical trajectory, comorbidities and right-heart findings reviewed alongside the calculator. These scores do not establish the diagnosis or automatically select treatment.

Assay, units and exact boundaries

Numeric blanks remain unknown. Nonfinite values, negative values outside the stated domain and unsupported functional classes prevent a result. BNP and NT-proBNP are alternative measurements of one component, not two votes; the selected assay determines its thresholds. ng/L and pg/mL are numerically equivalent, while pmol/L is not accepted.

Boundary rules remain specific to each published model. For example, a 440 m walk reaches the favorable REVEAL band but not the ESC >440 m band; BNP 800 and NT-proBNP 1100 also enter different bands in REVEAL and ESC. Original COMPERA retains the 2015 thresholds: BNP >300 or NT-proBNP >1400 ng/L is high risk; those cutoffs are not replaced by the later ESC/COMPERA 2.0 values. Available grades are rounded to the nearest integer, so means of 1.5 and 2.5 enter intermediate and high risk.

Original COMPERA treats cardiac index 2.0 to <2.5 as intermediate, preserving continuous values between the table's rounded 2.0–2.4 and ≥2.5 bands. Full REVEAL uses mean RAP >20 mmHg, DLCO <40% predicted and heart rate >96 beats/min. eGFR takes precedence over clinical renal-insufficiency judgment, which is used only when eGFR is unavailable.

Echo probability and incomplete findings

Echo interpretation uses measured peak TR velocity and the ESC supporting-sign categories: ventricles, pulmonary artery, and IVC/right atrium. Several findings within one category count once. Qualifying findings in at least two distinct categories affect probability; unchecked categories cannot silently become normal. A measured TRV above 3.4 m/s gives high echo probability independently of other signs.

The implementation distinguishes a jet explicitly reported as unmeasurable from a missing measurement. Neither an absent jet nor a low estimated pressure excludes PH. Severe TR, Doppler quality, RV dysfunction and the clinical phenotype require review. This tool interprets entered findings; it does not measure or analyze an uploaded echo image.

Haemodynamic phenotype and provisional grouping

PVR is calculated as (mPAP − PAWP)/CO using mmHg and L/min to produce Wood units. Cardiac index is not substituted for cardiac output. Resting thresholds use mPAP >20 mmHg, PAWP 15 mmHg and PVR 2 WU; a pre-capillary pattern alone does not establish group 1 PAH. Negative pressure gradients and material disagreement between entered and calculated PVR withhold interpretation.

Clinical grouping is a transparent rule-based differential across all five PH groups. Reported left-heart disease, CT parenchymal and vascular findings, V/Q pattern, PFTs and associated conditions add explicit supporting reasons and unresolved investigations. Competing mechanisms remain visible; the tool does not generate a patient-specific probability or a final diagnosis.

Mismatched V/Q defects prioritize expert evaluation for pulmonary vascular obstruction even after a negative CTPA. An isolated low DLCO does not assign group 3 or establish PAH. Spirometry and lung-volume cutoffs are teaching clues requiring laboratory lower limits of normal and clinical interpretation. Entry ranges and the numerical PVR conflict tolerance are original software safeguards, not new guideline diagnostic thresholds.

Ask the exPHert: management proposals

Ask the exPHert applies written rules to the entered findings and proposes a working clinical group even when the user supplies none. Its classification card separates clinician-provided information from proposed or unresolved classification, describes qualitative evidence fit and lists supporting findings, alternatives and missing investigations. Fit labels are not validated probabilities. A provisional group never silently checks a confirmation box or becomes an established diagnosis. It drives specific conditional named-drug proposals while the required phenotype, risk and prescribing prerequisites remain explicit. Reported urgent features remain visible independently of a risk score. Instability, unconfirmed clinical stability, a contraindicated PDE5-inhibitor/riociguat combination, pregnancy, invalid measurements, a possible PVOD pattern or discordant diagnostic findings withhold the routine PAH pathway.

Risk-linked active follow-up options require established adult group 1 PAH, current PAH treatment and a complete four-strata assessment. For a working PAH classification, the same entered variables can illustrate what the follow-up branch would be if adult PAH and the background regimen are confirmed; that conditional result is not presented as a validated prognosis for unclassified PH. The 2026 ERS guidance supplies context for discussion of add-on sotatercept above low risk, together with the continuing roles of parenteral prostacyclin and transplant assessment when indicated. Initial PAH options instead use a clinician-reviewed comprehensive baseline three-strata assessment with the comorbidity and vasoreactivity context. Low-risk and incomplete-data situations retain their separate contexts; no single score overrides severe right-heart findings or a conflicting diagnosis.

The therapy view presents named regimen examples with the reason for the pathway, eligibility conditions, sequencing and monitoring. Selected adult label doses are educational reference examples, not patient-specific starting orders; blood pressure, organ function, interactions, pregnancy assessment, formulation and local authorization remain prescribing prerequisites. It does not calculate an infusion dose or an individual titration plan. Uncertain current medication entries require reconciliation rather than being silently treated as absent. Working groups 2–5 receive conditional mechanism-specific options as well as investigation priorities. A suspected obstruction pathway can describe riociguat, anticoagulation and procedures as contingent future options; active CTEPH-specific proposals require a separately established indication, with acute PE, CTEPD without PH and other obstructions distinguished. Current US labels are identified as reference sources; they do not establish availability or approval in another jurisdiction. The clinician must verify the diagnosis, regimen, measurement quality and current guidance before acting.

Optional AI conversation

Ask the exPHert includes a separately password-protected free-text AI chat. Deliberately submitted questions and bounded conversation history are sent to OpenAI through the server; structured PH entries are not automatically included. A bounded selection from the curated module library supplies evidence context. Generated classification and therapy discussion remain unvalidated and require clinician verification. The structured guideline assistant stays a separate written-rule workflow. See AI chat: access, data flow & limits for the session, transmission and retention boundaries.

Current Evidence: qualitative benefit–risk

Current Evidence opens a separate public compact figure at /pulmonary-hypertension/current-evidence. Choose a clinical group before interpreting a drug or class. The default is group 1 in class view; the controls also show individual drugs and all five groups. Category order is fixed, with alphabetical ordering within each category, rather than an estimate of comparative efficacy.

The named categories are favorable, selected-context, mixed, uncertain and unfavorable. They are explicit editorial appraisals of a drug in a clinical group, with a rationale, population/route context, limitations and named supporting reports. They are not a computed risk ratio, hazard ratio, NNT/NNH, probability, prescribing score or formal GRADE assessment. Favorable does not mean harmless, uncertain does not mean no effect, and an unrated treatment is not assigned zero risk.

Appraisals activate only when their anchor reports are present with the expected tested-treatment attribution and population scope. Comparator-only, background-only, adjacent-population-only and unreviewed rows remain distinguishable rather than inheriting a favorable label. Report links permit inspection of the underlying endpoints and appraisal. New library records can expand the source index without acquiring a new clinical judgment automatically.

Class view retains member differences. Differing reviewed categories produce a mixed class summary, with the actual named member categories and their range disclosed. A favorable category with unreviewed members is labeled Favorable reviewed members and explicitly excludes those unreviewed drugs; their presence alone does not change the reviewed findings into a mixed result. The range describes differing editorial contexts, not a confidence interval, pooled class effect or an estimate for every member. Procedures and unrated treatments remain separate from rated drug comparisons.

Population and route matter within the same group. In group 3, inhaled treprostinil has a beneficial PH-ILD trial context in INCREASE, whereas PERFECT in PH-COPD stopped early for an unfavorable safety signal. Combining those contexts must retain the distinction; it cannot turn an ILD result into a COPD indication or extrapolate to oral, SC or IV administration. Other drug classes likewise retain phenotype, background-treatment and follow-up limits.

The trial index does not contain a standardized set of treatment/comparator benefit and harm event rates over comparable follow-up. The figure therefore does not divide favorable-report counts by harm-report counts, count overlapping cohorts as independent evidence, pool incompatible endpoints or calculate a clinical net-benefit ratio. A visually longer or differently placed marker encodes a category only; it is not a measured benefit magnitude or a head-to-head treatment ranking.

Cochrane interpretation guidance emphasizes patient-important outcomes, applicability and uncertainty; the FDA framework considers therapeutic context, evidence, uncertainties and risk management. These sources inform the qualitative approach but do not validate the application’s categories or graphic positions. Patient preferences, baseline risk, comorbidity, local authorization and clinical judgment remain outside this figure. It neither changes simulator drug-response coefficients nor supplies a patient-specific treatment recommendation.

Provenance and verification

Current Landscape is generated from the trial library and its explicit medication, clinical-group and primary-outcome metadata. All five PH groups remain visible and new indexed reports update the figure without a separate chart edit. Its markers count study reports, not pooled effects, treatment rankings or independent replications. Adjacent populations without uniformly RHC-confirmed PH, combined regimens, comparator or prior therapies, and safety appraisals are distinguished from primary findings attributed to the tested drug. Mixed-group reports retain their overall study result, not a claim of subgroup efficacy. Selecting a drug reveals the source-linked reports and can filter the evidence cards; the figure itself stays complete. Reports with incomplete metadata are visibly flagged rather than silently classified. Updating the indexed library is separate from automatically discovering or verifying new publications.

Trial cards and landscape study summaries identify administration separately for each indexed drug: oral, inhaled, intravenous, subcutaneous or procedural. The report-level route review is dated 31 October 2026 and checks full-paper methods, supplements and original study protocols, retaining formulation, infusion versus injection and transition sequences. An identified study formulation and its cited product instructions can establish a route when the paper does not repeat it; a drug name alone never determines the route. Where original full text is inaccessible, a study-specific published regimen table is distinguished from directly reviewed primary methods. Follow-up reports have explicit entries rather than automatic inheritance. Mixed cohorts retain their actual route combinations and any limits on participant-level attribution; missing verification remains explicit. Expanded cards explain formulation and regimen limits, while route wording is searchable. These labels describe the study intervention, not prescribing instructions or a claim that other routes have the same evidence.

Editorial applicability reviewed 31 October 2026 determines visual prominence, independently of outcome counts: established group-specific treatments lead, selected contexts retain their route or phenotype limits, and other studied treatments remain accessible through expansion or Show all. This is not a pooled efficacy ranking or prescribing advice. New indexed drugs default to other studied until explicitly reviewed. Group 2 emphasis identifies underlying heart-failure therapy, not a PH-specific indication. Group 3 inhaled treprostinil applies to PH-ILD, not PH-COPD or other routes; conditional PDE5-inhibitor consideration is limited to severe PH-ILD expert review with very low evidence. CTEPH drug options retain operability, residual-PH and anticoagulation context alongside PEA/BPA assessment. No universal PAH-drug approach is assigned to group 5.

The learning library distinguishes supplied teaching slides and papers from primary references newly consulted for the module. Historical slides do not take precedence over dated guidelines, trial reports or current prescribing information. Trial evidence and approved indications have different scopes; a cited drug study does not authorize treatment outside its population. Topline sponsor reports and uncontrolled extensions retain those labels. Randomized enrollment and smaller primary-analysis populations are distinguished where relevant. Cohort overlap is described rather than counted as independent replication. Registry plans, protocols and studies without verified efficacy results are not assigned positive or neutral outcome labels. The source review records held reports and unresolved denominators without filling gaps from secondary summaries.

Primary-outcome status is explicitly assigned independently of overall appraisal: met, not met, mixed primary findings, no confirmatory primary comparison, secondary report without a separate primary comparison, or uncertain. Nonconfirmatory reports include descriptive extensions and nonrandomized cohorts, some with observational comparators. Substudies, subgroup, secondary and pooled analyses use a separate secondary-report status instead of inheriting a parent trial result or implying that their findings are unverified. A met surrogate endpoint can coexist with an exploratory design, important harm or mixed benefit-risk; a neutral efficacy result is not relabeled as proven harm. Failed primary endpoints, harm signals and phenotype-specific limits remain visible alongside positive evidence. A composite hazard reduction is not relabeled as a mortality reduction; favorable secondary or subgroup findings do not rescue a failed primary comparison. The Real-world cohorts quick filter selects observational-context reports; open-label interventional trials are not relabeled as real-world evidence. For example, the trial library separates evidence for PDE5 inhibition in PAH from neutral or unfavorable trials in left-heart-associated PH. This distinction informs discussion and does not turn trial cards into an automated prescribing algorithm.

Boundary tests cover original COMPERA versus COMPERA 2.0 and ESC versus REVEAL thresholds, nearest-integer averaging, model score ranges, missing measurements, RHC recency, invalid entries and adult limits. Workup checks cover echo sign categories, exact haemodynamic cutoffs, invalid or conflicting inputs and overlapping clinical clues. Atlas checks compare preset sampling, cardiac-cycle means, resistance phenotypes and unchanged source physiology. Passing these checks establishes implementation consistency, not diagnostic accuracy, clinical efficacy or the validity of entering patient information.

Medication options to discuss is the explicit treatment-result heading. Each selected named regimen can carry an authored class label, such as Consider an ERA + a PDE5 inhibitor, alongside its drug examples. These labels are assigned only inside the existing conditional management branches, never inferred from a drug mention in free text. They do not change eligibility, risk or classification, and unknown-group, unstable, pregnancy, PVOD and conflicting-measurement restrictions remain visible. Established riociguat receives a coherent ERA/riociguat review option rather than a contradictory suggestion to add sildenafil or tadalafil.

Drug literature updates

The requested scope is all available PH pharmacotherapy evidence across groups 1–5, without an importance, sample-size, favorable-outcome, journal or design cutoff. A separate searchable source inventory retrieves complete documented PubMed, Europe PMC and ClinicalTrials.gov result sets. The coverage panel records each query, retrieval count, date and pagination completion. These overlapping source counts are not independent studies or patients. Search coverage still depends on terminology, indexing, database access and source availability; retrieval does not establish worldwide completeness.

The source inventory intentionally includes broad full-text matches, non-drug papers, animal work, reviews, protocols and tangential candidates. Explicit screening labels distinguish unscreened records, records ineligible for a PH patient drug-outcome appraisal, primary-source details that remain unavailable, and records awaiting further screening or detailed appraisal. A dated reason accompanies each recorded decision; records remain searchable and a decision is not an efficacy verdict. Exact PMID or registration matches link to the detailed appraisal when present; a shared registration may have multiple overlapping reports. Unlinked does not automatically mean excluded or unexamined. The detailed library separately retains explicitly reviewed population, denominator, comparator, endpoints, uncertainty and harms. When clinical-group assignment cannot be verified, the report keeps an explicit unresolved-group reason and does not acquire a group tag from the PH label alone. Ongoing and no-results registrations remain discoverable without receiving a positive or neutral outcome label.

Inventory search is public and returns bounded pages. Compressed bibliographic metadata stays on the server and is traversed as a stream rather than sending the entire corpus to the browser. No abstracts or full text are republished. Individually identified conference abstracts remain discoverable as metadata; an explicit abstract identity links an appraisal, while a shared proceedings DOI cannot appraise every abstract. Earlier records absent from a fresh query remain searchable with an earlier-snapshot note, not an inferred exclusion or retraction. The weekly watch refreshes complete source retrieval, checks new and updated indexing with an overlap window, expands phenotype/drug and historical citation searches, and works through the eligibility backlog without a relevance ranking or arbitrary inclusion quota. Failed or incomplete retrieval must retain its limitations and cannot advance a completed-search claim. When a live index changes during pagination, every observed candidate can be retained separately with partial, uncertified coverage, reported-count drift and cursor state. This does not replace the required complete broad source census. Curated exact source identities may link a verified preprint or earlier version to the published appraisal, without counting a new cohort; a shared DOI or an acronym alone does not establish that relationship.

The literature watch searches for new and previously omitted drug studies across all five clinical PH groups. Dated review records retain the actual search method, source-access limitations and unresolved candidates. Later indexing, online-first reports, corrections and final papers can change a record; shared DOI, registration or parent cohorts are checked before adding another report. A verified full publication can supersede a preliminary record without creating a second independent trial.

New classes remain mechanistically distinct: an sGC activator is not relabeled as an sGC stimulator, and early relaxin hemodynamic signals are read alongside randomized results. CTEPH anticoagulation evidence retains operative status, noninferiority margins and observational conflicts. Pediatric, thalassemia, dialysis, broadly selected HF and ILD cohorts retain their population limits. Adding a report does not automatically change treatment proposals, prescribing rules or simulator drug-response coefficients.