Moleculin ($MBRX): MIRACLE Part A Enrollment Closed At 89 Subjects, Comprehensive Readout Still Set For The First Quarter Of 2027

01 Bull, Base and Bear Scenarios After Data and Financing

September 14 scenario update: broader research may support additional development hypotheses, but the base case still rests on MIRACLE evidence and funded execution. The adverse case includes non-translatable preclinical findings, insufficient clinical benefit and financing delays or dilution. No success probability is raised solely because of conference acceptance.

What Would Falsify This Reading

The current assessment combines encouraging but unconfirmed clinical evidence with limited financial resources and a complex capital structure. These are separate hypotheses; new evidence can strengthen one while weakening another.

• Randomized efficacy: a materially smaller remission difference, unacceptable toxicity or uncertainty about dose selection in the full Part A analysis would weaken the development case. Consistent benefit with adequate follow-up would strengthen it. Subgroup or pooled blinded results cannot substitute for that comparison.
• Timing: a readout outside the guided first quarter of 2027 would require an updated operating and financing plan. The financial effect depends on the reason for delay, spending and any offsetting changes, rather than following automatically from the calendar.
• Financing: a committed transaction or partner agreement could reduce funding uncertainty. Its net cash, dilution, restrictions and obligations would need separate evaluation. Neither an unused ATM ceiling nor a hypothetical warrant exercise is committed funding.
• Listing: the September filing disclosed an intention to seek a hearing, not an approved compliance plan. An adverse Panel decision or conditions that cannot be met would worsen listing risk; a favorable determination would require its actual terms to be assessed.
• Share count: Series I warrants were repriced to $0.51. Crossing a strike does not compel holders to exercise. Reported exercises, new sales, ownership limits and contractual adjustments determine actual dilution.

Historical cash balances and accounting losses alone do not prove the exact date at which cash would run out. A fresh balance sheet or financing disclosure could materially change this assessment.

02 The First MIRACLE Unblinding: Randomized Clinical Numbers at Last

The June 30 analysis included 45 patients in the full intent-to-treat population: 14 in the Annamycin 190 mg/m² plus HiDAC arm, 14 in the 230 mg/m² plus HiDAC arm and 17 in the HiDAC-plus-placebo control. After one cycle, the 190 mg/m² arm reported 43% CR and 50% CRc; the 230 mg/m² arm reported 36% CR and 57% CRc; control reported 12% CR and 29% CRc.

The absolute CR advantage was therefore 31 percentage points for the 190 dose and 24 points for the 230 dose. For CRc, the advantage was 21 and 28 points. The most important feature is not the comparison with historical benchmarks, but the presence of a concurrent control treated under the same protocol, during the same period and after the same single cycle.

The lower dose produced the nominally higher CR rate, while the higher dose produced the higher CRc rate. That helps explain why the iDMC did not recommend dropping an arm: with only 14 patients per dose, the differences are too small and noisy to select the optimal dose reliably. Dose selection requires the complete efficacy, safety, pharmacokinetic and response-quality package from the completed Part A population.

03 Why “Positive” Does Not Yet Mean “Confirmed”

The iDMC unanimously concluded that there was a strong numerical trend favoring the experimental arms and sufficient evidence to continue. It did not declare statistically significant efficacy. That distinction is not cosmetic. MIRACLE uses an O’Brien-Fleming spending function that sets a very high threshold at early interim analyses and preserves nearly all alpha for the final analysis, designed for approximately 282 patients and 80% power to distinguish, under the protocol assumption, 20% CR in control from 35% CR in treatment.

An n=45 dataset is enough to test direction and plausibility, but not to close the registration debate. Investors still need remission duration, early mortality, infections, non-cardiac toxicity, subsequent transplant, response quality, molecular-risk distribution, performance after venetoclax and consistency between doses.

The statistical point: the lack of significance at the first interim is not a failure because significance was not required to continue. It cannot be turned into definitive proof either. The value of the readout depends on confirmation in the complete Part A dataset and then in the pivotal portion.

04 The July 31 Post-Venetoclax Update: Encouraging, but Still Blinded

At the July 18 data cut, 62 patients were evaluable. Preliminary blinded CR was 24% and CRc was 37%. Thirty patients, or 48% of the evaluable population, had failed a first-line venetoclax-based regimen; within that subgroup CR was 23% and CRc was 37%, essentially identical to the overall population.

The stability of the signal is notable because the post-venetoclax share increased from 31.1% in the n=45 population to 48% in the n=62 dataset. In other words, the study absorbed a harder patient mix without blinded CRc falling outside the approximately 37-40% range seen across sequential analyses.

The company compares this with a published retrospective cohort in which only 13% of patients receiving salvage therapy after venetoclax plus a hypomethylating agent achieved CR/CRi and median overall survival was 2.4 months. The comparison is useful context, but not controlled: the historical cohort was small, single-center, heterogeneous and different from the MIRACLE protocol.

Prudent interpretation: the post-venetoclax data strengthen the plausibility that Annamycin retains activity in a difficult population. They do not yet show how much of the remission activity belongs to Annamycin because the 62-patient analysis remains blinded and includes control.

05 Enrollment and the Q1 2027 Part A window

Enrollment closure changes the next task from recruitment to treatment completion, follow-up, data reconciliation and analysis. The guided release is a window; the actual date and completeness of the dataset still need confirmation.

06 MIRACLE Trial Design: Why Part A Is More Than Ordinary Dose-Finding

MIRACLE is a global, multicenter, randomized, double-blind, placebo-controlled adaptive study. Part A compares Annamycin 190 mg/m² plus HiDAC, Annamycin 230 mg/m² plus HiDAC and HiDAC plus placebo. Part B is expected to carry one optimal dose forward against control, with Part A data carried into the pivotal efficacy analysis.

This structure gives value to data already collected, but increases the importance of dose selection. A higher dose is not automatically better; it must offer the best balance of CR, CRc, safety, pharmacokinetics and treatment completion. Both arms beating control is positive. Neither clearly dominating the other makes the expanded Part A dataset important.

The trial enrolls adults in second line after one prior induction therapy. As of July 31, the company reported patients from seven countries and 33 sites that had undergone site-initiation visits, with a target of at least 45 sites for Part B. European recruitment has contributed more than U.S. recruitment, which Moleculin said it intends to improve.

07 Annamycin and Cardiac Safety: Real Differentiation, but Not a Complete Answer

Anthracyclines are powerful but limited by cumulative cardiotoxicity. Annamycin was designed with a multilamellar lipid delivery system to improve distribution, bypass multidrug resistance and reduce cardiac injury. The ASCO 2026 poster reported no detectable cardiotoxicity in pooled completed-study data, supported by independent cardiac review, ejection-fraction assessments, ECGs, biomarkers and other available measures. A pooled analysis of 90 patients from five trials has since been published peer-reviewed in Frontiers in Cardiovascular Medicine (September 3, 2026; publicized September 28), with the limits of a small, non-randomized sample and short follow-up.

On July 31, the company reiterated that no evidence of cardiotoxicity had emerged in MIRACLE based on reported ejection fractions and adverse events. That matters because Annamycin’s clinical value is not simply producing remissions, but doing so with a historically effective drug class without its defining safety limitation.

The correct wording remains “no signal observed to date,” not “cardiac risk eliminated.” The dataset is still limited, follow-up continues and registration will require a complete safety assessment, not cardiac safety alone.

The pooled publication also reports five patients unable to continue after allergic reactions to Annamycin or cytarabine and two withdrawals after clinical deterioration. The median interval between paired echocardiograms was 53 days. Thus, a reassuring cardiac analysis is not equivalent to an absence of adverse events or proof of long-term safety. Two transient ECG findings were not classified as unrelated, although investigators did not deem the reported ECG changes serious. These distinctions matter when interpreting the broader cardiac conclusion.

08 The completed August offering and its economic trade-off

The July 31 offering was completed on August 3, according to Note 8 of the June-quarter filing. It issued 11,088,334 common shares and pre-funded warrants for another 1,288,333 shares, together with Series I warrants covering 37,130,001 shares. The pre-funded warrants were exercised shortly after closing. The combined offering price was $0.75 per common share or pre-funded equivalent with its associated warrants. Gross proceeds were approximately $9.3 million and expenses approximately $0.8 million, implying about $8.5 million net from those rounded disclosures. Form 10-Q, subsequent events.

The distinction between pricing and closing matters. The earlier announcement described a best-efforts transaction subject to completion; the later filing supplies confirmation of securities issued and proceeds received. It does not establish how much of that cash remained at September-end. Clinical expenditure continued between the balance-sheet date, closing and the time of writing.

Each unit included three Series I warrants. Their original $0.75 strike was reduced to $0.51 after September ATM sales, and the disclosed contractual floor is $0.21. They expire on the fifth anniversary of the August 3 closing. Full cash exercise of all 37,130,001 at $0.51 would yield $18.936 million gross and create that many new shares. This is an illustration, not a forecast or a financing commitment; holder decisions, ownership limitations, registration provisions and any further amendments matter.

The economic trade-off is therefore more specific than saying the offering was either good or bad. Cash supports continued operations, while new common shares reduce prior holders’ percentage ownership. Warrants create additional potential dilution and may supply future cash if exercised for cash. A lower strike increases the number of shares obtainable for each dollar of exercise proceeds compared with the original terms, without itself issuing those shares.

The July 31 price decline occurred on the same day as the financing and blinded clinical update. Timing alone cannot establish how much of the move was caused by either announcement, market conditions or other trading. It is not evidence that the clinical outcome is known. The relevant documents establish the funding terms and clinical observations; they do not identify the motives of every buyer and seller.

09 Capital structure and the October shareholder vote

The August 19 proxy snapshot reported 19,477,380 outstanding common shares plus 48,290,275 underlying warrants, 199,854 underlying options and 73,614 underlying unvested stock awards. Adding those categories produces 68,041,123 common or potential shares, an illustrative total rather than issued stock or a forecast. The proxy’s dated categories are preferable to combining older pre-offering warrant balances with later issuance without accounting for intervening exercises and adjustments. Proxy capital-structure table.

By September 23 the company reported 20,385,780 common shares after issuing 888,400 ATM shares during September for $0.48 million gross. The difference from the August 6 share count is 908,400, so those disclosed ATM sales alone do not explain every share in the change. The September filing does not explain the remaining 20,000. September 23 Form 8-K.

Series I anti-dilution provisions reduce the exercise price upon qualifying below-strike issuance, subject to exempt transactions and a $0.21 floor. The September sale reset the strike from $0.75 to $0.51. Note 8 of the June-quarter filing also states that the August offering already adjusted Series G and H exercise prices to $1.326 and $0.962 respectively. Those adjustments are disclosed events, not hypothetical future repricings. Subsequent-events note.

The August 21 prospectus registered up to $19.7 million additional gross ATM sales, excluding prior sales. Its baby-shelf public-float calculation used the June 29 price of $3.57, the highest close in the preceding sixty days, and nonaffiliate holdings to arrive at approximately $64.9 million. That regulatory calculation is different from a September market capitalization. The remaining usable capacity depends on sales already made and applicable limits; the headline registration amount is neither current cash nor a promise that all shares can be sold on favorable terms.

The October 9 meeting approved authority for a reverse split between 1-for-2 and 1-for-20 at the board’s discretion within a year. Authorized common shares remain 500 million under the approved proposal. Authorization and implementation are separate events: this filing reports the vote, not an implemented split. A reverse split does not inject cash or demonstrate operating improvement. The equity-plan increase of 3,861,894 shares was approved, bringing its reserve to 3,875,999; the proposal to remove supermajority voting requirements was not approved. October 9 Form 8-K.

An illustrative financing sensitivity shows why net proceeds and share count need to be considered together. Raising $5 million gross at $0.50 would require 10 million shares before expenses or accompanying securities, equivalent to 49.1% of the September reported count. At $0.40 it would require 12.5 million, or 61.3%. Existing holders’ collective ownership would then be about 67.1% or 62.0% of the respective post-issue totals, absent other changes. These are arithmetic examples, not proposed terms or assumptions that the company will raise at either price.

10 Financial statements: cash use, accounting losses and subsequent proceeds

The June 30 balance sheet reported $7.262 million cash, $8.723 million current assets and $9.313 million current liabilities. The resulting working-capital deficit was $0.590 million. Total assets were $21.087 million, including $11.148 million intangible assets; total liabilities were $9.500 million and equity $11.587 million. Intangibles are not cash available to pay trial invoices. These are June balances, before the completed August financing. June financial statements.

For the second quarter, revenue was zero, R&D $5.450 million, G&A $2.103 million and total operating expenses $7.561 million. Net loss was $7.635 million. First-half operating expenses were $15.435 million and net loss $20.480 million. The income statement included a $10.763 million gain from warrant remeasurement and a $15.158 million loss on warrant issuance, explaining why net loss is not a clean measure of operating cash consumption.

First-half cash used in operations was $12.768 million, versus $10.146 million a year earlier. Financing supplied $11.225 million: $7.608 million net from securities transactions and warrant inducement, $2.817 million from warrant exercises and $0.800 million net from ATM issuance. With no investing cash flow and a negative $0.073 million currency effect, cash declined $1.616 million from $8.878 million to $7.262 million. This bridge uses reported cash-flow categories rather than treating noncash warrant accounting as an additional funding requirement.

Operating cash use averaged $2.128 million a month over those six months. Dividing June cash by that historical rate gives approximately 3.41 months, but this is not an October runway estimate: it omits subsequent financing, later spending and any change in activity. Similarly, adding the August net proceeds to June cash without subtracting intervening cash use would not produce a current balance. The August results release attributed runway into Q1 2027 to the company’s then-current plan; September’s update still warned that significant additional financing was needed and uncommitted.

The difference between accounting expense and cash use also affects comparisons across quarters. Accounts payable rose by $0.829 million and accrued liabilities by $1.630 million in the cash-flow reconciliation, temporarily supporting cash relative to recognized expenses. Those movements do not mean the underlying obligations disappeared. A later payment cycle can consume cash even if reported operating expense is stable.

As an explicitly illustrative sensitivity, an additional three months at the first-half average operating cash-use rate would require $6.384 million; at 25% higher spending it would require $7.980 million. These calculations show the scale of schedule exposure, not a prediction of delay or the amount management must raise. Actual needs depend on trial work, payment timing, other programs, financing costs and any changes to the development plan.

11 Market positioning and the limits of sentiment

The dated September 30 market snapshot records a $0.50 share price and approximately $10.2 million equity capitalization using 20,385,780 common shares. That calculation excludes unissued warrant and option shares. Combining it mechanically with a June cash figure does not yield the market value of the drug as a residual.

Third-party short interest, institutional holdings and insider percentages can differ in definitions, reporting delays and denominators. They are positioning inputs rather than a contemporaneous register. In particular, a short-interest percentage can change because short shares, float or both change; the percentage alone does not identify covering or new bearish positions.

Retail posts are anecdotal observations from self-selected participants. They do not establish the quality of a clinical endpoint, financing capacity or the motives of the broader market. Older August posts about this company do not indicate current sentiment.

12 Company Overview and Pipeline Beyond MIRACLE

Intellectual property: a Mexican patent to 2040

On September 23, 2026 Moleculin announced that the Mexican Patent Office granted Patent No. 436849, covering its lung-targeted liposomal Annamycin formulation, designed to improve delivery to pulmonary tumours and metastatic disease. The company gives an expiry of November 23, 2040 and describes the grant as part of an international patent estate across multiple jurisdictions, without naming them in the announcement. A national patent protects the invention in that market; it is not a regulatory authorisation, a clinical result or a commercial agreement. Moleculin announcement · Form 8-K

Annamycin beyond AML: September research update

On September 25, Moleculin reported preclinical pancreatic-cancer data suggesting that CD8+ T cells contribute partly to Annamycin activity. This follows the September 14 schedule of presentations on tissue distribution, topoisomerase II and pancreatic models. Annamycin is not currently in clinical development for pancreatic cancer. These findings are not patient-response or survival data and are separate from the MIRACLE AML study.

Moleculin is headquartered in Houston and develops oncology and antiviral candidates. Annamycin is the principal clinical program. Its soft-tissue-sarcoma research and possible future external studies address distinct indications; the September pancreatic announcement explicitly described preclinical work, not an active pancreatic trial. WP1066 targets p-STAT3 and related signaling, while WP1122 belongs to the metabolism-focused portfolio. Each program requires its own evidence and funding; no compensating value is assumed if MIRACLE disappoints.

13 AML Competitive Landscape: The Advantage Must Be Clinically Usable

R/R AML is fragmented by mutations, age, fitness, prior treatment and transplant eligibility. Targeted therapies exist for FLT3, IDH1, IDH2 and menin-related subgroups, alongside venetoclax combinations, hypomethylating agents, cytarabine-based regimens and experimental immune strategies. Annamycin does not need to prove there are no alternatives; it must demonstrate a credible advantage in a broad biomarker-agnostic population.

The potential positioning is an agent capable of producing deep remissions after one cycle, retaining activity after venetoclax failure, serving as a bridge to transplant and reducing the cardiac limitation of conventional anthracyclines. A strong clinical proposition also requires response duration, survival, complete safety, multicenter feasibility and a regulatory path aligned across the FDA and EMA.

Existing targeted approvals address defined populations; for example, FDA: revumenib in susceptible NPM1-mutated relapsed/refractory AML. This is context for segmentation, not a head-to-head comparison with Annamycin.

14 Management, ownership and incentives

Walter V. Klemp is chairman, chief executive and president; Jonathan P. Foster is chief financial officer. Management’s development judgment and financing decisions both matter to common shareholders. Holding equity can align interests, but it cannot establish clinical success or make the terms of future funding immaterial.

Klemp’s August 25 Schedule 13D reports 2,070,761 beneficial shares, rounded to 9.9%, on an August 19 event date. The aggregate comprises 564,976 common shares, including 636 held through AnnaMed, 1,250,959 Series I warrant shares counted under the ownership limitation, and 254,826 option and other warrant shares exercisable within sixty days. This is holder-specific beneficial ownership, not the company’s fully diluted share count. The calculation adds relevant exercisable instruments to the applicable denominator.

The filing explicitly describes an investment in the August 3 offering: Klemp purchased 533,333 common shares with 1,599,999 accompanying Series I warrants for approximately $400,000 using personal funds. It also describes compensatory securities separately. It would therefore be incorrect to classify the entire beneficial aggregate as compensation or as common shares already held. Other warrant shares are excluded because of exercise limitations. Klemp Schedule 13D.

Foster’s separate August 25 filing reports 6.3% beneficial ownership. That percentage reflects its event date and exercise assumptions; it is not a contemporaneous ownership register for October. The two percentages are not additive with the insider ownership reported by data providers, which can use a different definition and date. A later change in a third-party data field cannot by itself establish an insider purchase, sale or absence of trading. Foster Schedule 13D.

15 Analysts and the Valuation Framework After the Offering

Any price target published before July 31 must be rebuilt around the new capital structure. In a nano-cap with reverse splits, warrant inducements, new shares and anti-dilution instruments, a nominal target is not useful without fully diluted share count, capital required through registration and a risk-adjusted probability of success.

The more useful framework is a sequence of questions: which dose will be selected? Will the CR advantage remain close to the first interim? Will control stay near 12% or rise? Will the post-venetoclax CRc signal be confirmed unblinded? How much cash is needed for Part B? How many warrants become economically exercisable? Is a partner willing to fund the program?

16 Catalyst calendar and evidence required

Enrollment closure and the August financing are completed historical events. The September pancreatic presentation and cardiac-safety publication are also already disclosed. They are no longer future calendar items, even if older versions of a schedule listed them.

As of October 4, the registry record, last posted August 5, still lists MIRACLE recruiting, an estimated 312 participants, estimated primary completion in August 2029 and overall completion in August 2030. The June issuer statistical explanation referred to approximately 282 subjects in its final analysis. These are different disclosed planning figures; neither is the actual Part A enrollment count. The September company announcement is more recent on Part A closure, while the registry describes the overall study. Its estimated completion dates are not scheduled release dates for the Q1 2027 Part A analysis. MIRACLE registry.

17 Red Flags That Cannot Be Ignored

• Sample size: the first unblinding contains only 14 patients per dose and 17 in control.
• No statistical significance: consistent with the interim design, but the advantage must be confirmed.
• Early endpoint: remission after one cycle does not replace duration, survival and full clinical quality.
• Dilution: new shares/PFWs exceed the pre-offering share base; new warrants are more than five times that base.
• Warrant repricing: existing instruments can move closer to market through anti-dilution floors.
• Runway remains limited: even after the offering, Moleculin expects additional funding needs before registration.
• Nasdaq compliance: the company had bid-price and equity issues in 2025, and on September 15, 2026 Nasdaq issued a delisting determination with no 180-day cure period; a timely hearing request stays delisting pending the Panel decision.
• Execution: Part B requires sites, capital, dose selection, regulatory work and required nonclinical completion.

The research broadens the questions being studied beyond AML. It does not resolve the randomized clinical question or the financing requirement. Moleculin reiterates that significant additional financing is needed and that it has no commitments for that financing; the described milestones depend on obtaining it. The earlier 45-patient MIRACLE results did not achieve statistical significance and may not predict the completed trial.

Evidence limit: the September 25 company release describes preclinical pancreatic models, not human efficacy. A possible immune contribution does not validate MIRACLE, demonstrate a survival benefit in patients or remove the funding requirement.