On September 18, 2026, FDA issued Warning Letter 738238 to Empower Clinic Services, LLC, doing business as Empower Pharmacy, after a November 3-14, 2025 inspection of the firm’s Houston 503A pharmacy at 7601 N. Sam Houston Parkway West. The letter was posted to FDA.gov on September 22.
By the end of the week, KING 5, Buchanan Ingersoll, MedPage Today, and the Partnership for Safe Medicines had all written about it. Most of that coverage starts with sterility. The sterile-process observations are real, and they belong in the same PDF. They are not the first finding.
The letter’s opening allegation is that products Empower compounded failed the conditions of section 503A of the Federal Food, Drug, and Cosmetic Act. FDA named the formulations: tirzepatide with niacinamide, and semaglutide with cyanocobalamin. The agency’s position is that those products “appear to be essentially copies of FDA-approved semaglutide and tirzepatide products,” compounded “regularly or in inordinate amounts.”
Here is what makes this letter analytically useful rather than simply another enforcement headline. A warning letter is a record. It can be read the same way a Certificate of Analysis is read: named products, named methods, named lots, and a clear line between what the paper claims and what the paper can support.
At Vanguard Laboratory we do not inspect pharmacies, and we did not test these lots. We read analytical and regulatory records for a living. Vanguard does not take a position here on prescribing, clinical use, or whether any particular compounded GLP-1 should be on a clinic’s shelf. The question is narrower. What does this letter actually say, and what should a clinic, a med spa, or another compounding pharmacy ask to see before the next vial goes into a refrigerator.
Two Halves of the Same Letter
FDA’s compounding-actions page already lists prior Empower warning letters dated October 15, 2021 and April 2, 2025. The September 18 letter is about the 503A pharmacy inspected in November 2025. FDA acknowledged written responses dated December 8, 2025, March 18, 2026, and April 30, 2026, then issued the warning letter anyway.
Empower’s response, through spokesperson Emma Neal in a KING 5 report dated September 24, is that the firm has “already remediated many of the issues outlined in the Warning Letter” and will continue working with the agency. A warning letter is not a court judgment. It is FDA’s enforcement position, on a public docket, with a fifteen-working-day clock to respond.
Read past the headlines and the letter splits into two findings that should not be collapsed.
The 503A half is about copies. The question FDA is asking is whether adding a vitamin, generating a “significant difference” sentence, and filling a large number of the same SKU is compounding for an identified individual patient, or producing an analog of a commercial drug under a compounding exemption.
The sterile-process half is about whether the ISO 5 environment was ever shown to protect product under real filling conditions, and whether a later media-fill summary can be reconciled to the vials it claims to describe.
Those halves have different legal consequences. Mixing them is how a useful document turns into a general quality complaint.
What 503A Actually Requires
Section 503A is a narrow exemption, not a manufacturing license. A licensed pharmacist in a state-licensed pharmacy may compound a drug for an identified individual patient based on a valid prescription. If the rest of the statute’s conditions are met, three requirements drop away: current good manufacturing practice under section 501(a)(2)(B), labeling with adequate directions for use under section 502(f)(1), and FDA approval under section 505.
That is the commercial logic of 503A. Compounding is supposed to be a patient-specific exception to industrial drug manufacturing. Stay inside the exception and the pharmacy is not treated as a drug manufacturer.
One of the conditions is easy to quote and hard to document. The pharmacist “does not compound regularly or in inordinate amounts any drug products that are essentially copies of a commercially available drug product.”
The statute then defines the exception to that restriction. A product is not essentially a copy if “there is a change, made for an identified individual patient, which produces for that patient a significant difference, as determined by the prescribing practitioner, between the compounded drug and the comparable commercially available drug product.”
Two other conditions almost never make the news, and they are the ones that belong in a testing lab’s week.
Bulk drug substances used in 503A compounding must be “accompanied by valid certificates of analysis for each bulk drug substance.” That is 21 U.S.C. § 353a(b)(1)(A)(iii). It is not a Vanguard preference. It is in the compounding statute.
And the compounding is for an identified patient. Not a catalog. Not a telehealth menu. A person.
Lose those conditions and the exemptions vanish. FDA’s letter is explicit about the consequence. The products it treats as 503A-ineligible are, in the agency’s view, unapproved new drugs, misbranded for lacking adequate directions for use, and subject to CGMP. Introducing them into interstate commerce is a prohibited act.
That is a change in legal category, not a documentation note.
The Formulations FDA Named
FDA did not speak in generalities. Investigators collected evidence on three SKUs:
- Tirzepatide/niacinamide, 4 mL, 17/2 mg/mL
- Semaglutide/cyanocobalamin, 1 mL, 5/0.5 mg/mL
- Tirzepatide/niacinamide, 2.5 mL, 8/2 mg/mL
The agency’s sentence: these “appear to be essentially copies of FDA-approved semaglutide and tirzepatide products.”
FDA previously exercised enforcement discretion for state-licensed pharmacies compounding semaglutide and tirzepatide injection products that were essentially copies of commercially available drugs. That discretion ended for tirzepatide injection on March 5, 2025, and for semaglutide injection on April 24, 2025. The production months cited in the letter, July through October 2025, sit after those dates.
The monthly order counts are redacted as (b)(4). We are not going to invent them. FDA’s unredacted claim is the one that matters: the firm compounded and filled “more than (b)(4) orders” of each named product in each of those four months, and “the volume of products you are producing suggests that differences between products you are compounding and the FDA-approved products are pretextual.”
Volume is not, by itself, a violation. Buchanan Ingersoll’s September 25 analysis of the same letter is careful on this point, and the care is warranted. High volume does not automatically eliminate the statutory significant-difference provision. What volume does, in FDA’s hands, is raise the question of whether the asserted difference is real.
A Vitamin Is Not, By Itself, a Significant Difference
The added ingredients are not a secret. Niacinamide. Cyanocobalamin. B3 and B12, written onto a GLP-1 label.
FDA does not say those molecules are forbidden. It says the record did not show they were prescribed as a change that produced a significant difference for an identified patient.
Buchanan Ingersoll, reading the same letter, put it this way: adding niacinamide, vitamin B12, or another ingredient does not, standing alone, establish that the resulting product sits outside the essentially-copy restriction. The analysis involves both the formulation change and the prescriber’s determination concerning the particular patient.
What FDA collected, in three buckets:
First, orders and prescriptions that lack any prescriber determination of significant difference.
Second, orders and prescriptions with purported determinations “repeated verbatim across many records,” which “may be pre-generated for selection by the prescriber, rather than written by the prescriber for an identified individual patient.”
Third, production records showing regular compounding or inordinate amounts.
Then this sentence, which will outlive the SKUs:
“Generating prescriptions through means that undermine the individualized nature of a prescriber’s clinical judgment (for example through third-party technology platforms that provide prescribers with pre-selected menu options for choosing a statement of significant difference) call the individualized nature of those determinations into question, potentially undermining claims that such prescriber determinations of significant difference are sufficient to meet the conditions of section 503A.”
FDA is not banning software. It is saying a menu can strip the determination of the one quality the statute requires: that it belong to a patient.
A clinic or a sister pharmacy that wants a defensible file does not need to litigate Empower to use this. The operational test is smaller.
Can you produce the patient-specific determination, in language that belongs to that patient, for the lot in the refrigerator. If the answer is a checkbox that reads the same on every prescription, FDA has now said what it thinks that checkbox is worth.
Insanitary Conditions Do Not Wait for 503A to Fail
The sterile-process observations sit under a different statute.
Section 501(a)(2)(A), the insanitary-conditions provision, applies whether or not a product qualifies for 503A exemptions. FDA wrote that into the letter so it could not be missed: “Please be aware that section 501(a)(2)(A) of the FD&C Act concerning insanitary conditions applies regardless of whether drug products you compound meet the conditions of section 503A.”
The firm failed to perform adequate smoke studies under dynamic conditions to demonstrate unidirectional airflow within the ISO 5 area. A smoke study is how a sterile operation shows that air in the critical zone still moves in one direction when people are filling, not when the room is empty. A static demonstration is a photograph of a process, not a test of one.
Media fills were not performed under the most challenging or stressful conditions. A media fill is a process simulation. Growth media replaces product, the line is run, and vials are incubated to see whether the process can be executed without contamination. If the simulation is easier than production, it does not simulate production. FDA’s conclusion: “there is a lack of assurance that your firm can aseptically produce drug products within your facility.”
For the products FDA treats as 503A-ineligible, the letter then recites CGMP, including 21 CFR 211.22 (quality-control unit: identity, strength, quality, and purity), 211.113(b) (aseptic-process validation), 211.42 (ISO 5 equipment and environmental monitoring), and 211.100(a) (production controls). Those citations attach to the ineligible products. The insanitary-conditions observations do not need them. A pharmacy cannot close this letter by rewriting GLP-1 prescription templates and leaving the ISO 5 record alone.
The Summary Said Zero. The Vial Count Did Not Match.
The most lab-shaped paragraph in the letter is not about airflow. It is about a document.
Empower executed an aseptic media-fill simulation in January 2026. The summary FDA reviewed stated that after a 14-day incubation, all vials were visually inspected for turbidity or microbial growth, and “no contamination was observed in any of the (b)(4) vials, confirming the sterility assurance of the process.”
FDA compared that sentence to the underlying records and could not make the numbers meet.
The summary “does not appear to match the number of vials inspected per the documents provided.” Based on the vials recorded, “it appears up to (b)(4) vials were examined.” No explanation was provided for the discrepancy. Records documented results for some units “with no documentation or explanation” for others. Multiple lines were labeled “N/A.” FDA noted that “N/A” was not defined. It is unknown, the letter says, what the designation means.
Some original results had been transcribed onto Form Revision 3 because they were documented on the wrong form. Transcribing data onto a different document, FDA wrote, “even if done for legitimate administrative reasons, must be carefully controlled, documented, and justified. Without a clear audit trail showing why results were rewritten and who authorized it, this practice raises concerns about whether the data accurately reflects what actually occurred during the media fill.”
That is a data-integrity finding. In a testing laboratory, a result that cannot be traced to the vials is not a result. It is a claim formatted to look like one.
A later smoke study, under Revision 005, was executed March 30 through April 3, 2026. Empower’s response described it as remaining “in controlled post-execution review.” FDA had not been given a final formal summary report, and could not fully evaluate the corrective action.
None of this requires a theory about what grew or did not grow. It requires the same habit we ask of every CoA that comes across the bench. A number without a reconciliation is not a number.
The Testing Requirement Already in the Statute
Section 503A already requires that bulk drug substances be accompanied by valid certificates of analysis. FDA’s letter then reminds the firm of a second, related point: if you contract a laboratory to perform CGMP functions, you still own the result.
“If you choose to contract with a laboratory to perform some functions required by CGMP, it is essential that you select a qualified contractor and that you maintain sufficient oversight of the contractor’s operations to ensure that it is fully CGMP compliant. Regardless of whether you rely on a contract facility, you are responsible for assuring that drugs you produce are neither adulterated nor misbranded.”
Sending the sample out does not send the signature with it. You cannot deputize quality. You can only document it.
What to Ask Before the Next Vial
A warning letter is not a CoA. A CoA is not a warning letter. They answer different questions, and you need both kinds of record.
For the legal basis of a compounded analog of a commercial drug:
Is this patient-specific compounding, or a high-volume analog of a commercial product. Where is the prescriber’s significant-difference determination for this patient, in this patient’s language. If a platform generated the sentence, can anyone show it was the prescriber’s determination rather than a pre-selected menu option.
For the vial itself:
Identity of the active, by LC-MS or an equivalent orthogonal method, lot-matched to the container in hand. Strength or assay as a number with units, not the word “pass.” Purity by HPLC, with the method named. Bacterial endotoxin as a quantitative value with a named method. Sterility with the method and the incubation duration named. And the bulk drug substance CoA that 503A already requires.
If those elements are missing, the certificate documents that a test was mentioned. It does not document that the product is what the label says, at the strength the label says, in a container that was filled under control.
The Vanguard Standard
Vanguard Laboratory is ISO/IEC 17025:2017 accredited, follows AOAC and USP methodology, and runs instruments 24 hours a day, seven days a week. Identity by LC-MS. Purity by HPLC-UV. Potency by quantitative assay. Safety by ICP-MS for heavy metals and LAL or recombinant methods for bacterial endotoxin. We report results on a clearly stated basis, because a number without a basis is not a result.
We did not test these Empower lots. We are not going to treat a public warning letter as a substitute for a lot-matched certificate, or a lot-matched certificate as a substitute for a pharmacy’s own sterile-process qualification. The paper and the vial have to describe the same product. When they do not, the gap is the finding.
FDA told Empower to take prompt action, and reserved seizure and injunction. That is the agency’s next move, if it makes one. It is not ours.
Ours is the same as it was before September 18. Measure what is in the vial. Name the method. Make the numbers reconcile.
Need identity, purity, potency, or endotoxin testing on a specific lot? Contact Vanguard Laboratory at [email protected] or call 360.967.7010.
Sources
- U.S. Food and Drug Administration. “Empower Clinic Services, LLC dba Empower Pharmacy MARCS-CMS 738238.” Warning Letter, September 18, 2026. https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/warning-letters/empower-clinic-services-llc-dba-empower-pharmacy-738238-09182026
- 21 U.S.C. § 353a. Pharmacy compounding (section 503A of the FD&C Act).
- Buchanan Ingersoll & Rooney PC. “FDA Warning Letter to Empower Pharmacy and What It Means for 503A Compounding Pharmacies.” September 25, 2026.
- KING 5 Investigators. “FDA warns compounding pharmacy over sterility of GLP-1 drugs.” September 24, 2026.
- Partnership for Safe Medicines. “Statement on FDA’s Warning Letter to Empower Pharmacy for producing drug products in violation of federal law.” Shabbir Imber Safdar, September 23, 2026.
- U.S. Food and Drug Administration. “Compounding: Inspections, Recalls, and other Actions.” Empower Clinic Services, LLC entries (Warning Letter 10/15/2021; Warning Letter 04/02/2025).