Peptide Research Literature Index
A structured, dated index of the research-peptide landscape

Why do most of these cases end in probation? Eight peptide prosecutions ran through the same guideline, and the line that decides between no prison and thirty months is whether anybody is found to have lost a dollar

Reference page · published 2026-10-26

Disclosure: this index is operated by Artemis Labs, which sells research compounds to laboratories. We have a direct interest in this page and it cuts both ways, so it goes at the top, not the bottom: two of the compounds named in the cases below — PT-141 (bremelanotide) and melanotan II — are compounds Artemis Labs sells. Artemis Labs has no connection to any person or company named on this page. Nothing here is legal advice. Every quotation below is verbatim from a federal court filing, a federal judgment, a federal docket entry, a Department of Justice press release, the United States Code, or the United States Sentencing Commission’s Guidelines Manual — each one fetched and read for this page on October 26, 2026, and each listed under Sources. Where we could not verify something, the page says so instead of filling the gap.

Short answer: the sentence turns on whether a court finds that somebody lost money. Misbranding cases that involve fraud are sentenced under the fraud guideline, § 2B1.1, and the first thing that guideline does is add offense levels for loss. In four of the cases below the loss was found or agreed to be nothing, the offense level came out at 4, and the Sentencing Table reads 0–6 months — a range in which probation is expressly authorized. In two others, a special rule in the same guideline made the loss equal to what customers paid, which added 14 levels and produced a range of 30 to 37 months. That is a gap of fifteen levels, and nothing else in the calculation comes close to it. The argument over that one number is the whole sentence — and the lawyer who built the no-loss argument lost it for his own client, eighteen days after he filed it.

The question, and why court records rarely answer it

Read enough federal peptide prosecutions and a pattern shows up that nobody explains. Most of them end without prison. A handful end in two or three years. The cases look alike from the outside: same statute, same kind of website, same disclaimer, often the same defense lawyer.

The usual explanation offered is that the compound mattered, or that the government went harder on the bigger sellers. Neither survives contact with the filings. One of the defendants below moved $3.2 million in revenues and served ten months. Another moved less than a quarter of that and served three years.

The real variable is a line in a sentencing table. This page shows where it comes from, who argued about it, and what happened each time one side won. We have published the single-case version of this before — one prosecution where both sides argued the loss question on the record. What is new here is the other side of that argument, and the eight cases where it won.

Start with two sentences that should not be different

Two defendants. Both sold research chemicals through websites carrying a research-use disclaimer. Both pleaded guilty. Both had no meaningful criminal record. Here is the money and the outcome.

 Revenue in the recordLoss figure usedOffense levelSentence
The case filed as Research Chemical PSR (defendant redacted)over $670,000 in revenuesnone — stipulated4three years’ probation, two months’ home confinement
United States v. Ash (W.D. Pa.)greater than $550,000 but not more than $1,500,000the revenue1924 months in prison

The first defendant made more money than the second. The first got no prison. The difference is not a judgment about who was worse. It is that in the first case the parties agreed the loss was nothing, and in the second they agreed it was the revenue.

The first case comes to us only as an exhibit, with the defendant’s name removed because presentence reports are confidential. The brief that attached it describes the stipulation this way: it was stipulated that there was no loss on the same facts as here, and that Probation adopted the analysis despite over $670,000 in revenues. The same brief says a sentencing transcript in that case records the prosecutor’s own explanation — that the significant factor behind the absence of a loss enhancement was that there was no consumer that was defrauded.

How a misbranding sentence is actually built

Federal sentencing starts with arithmetic. The court works out an offense level, reads it against the defendant’s criminal history, and looks up a range of months in a table. The range is advisory — a judge may sentence above or below it — but every sentencing starts there.

Food-and-drug offenses have their own guideline, § 2N2.1. We pulled it from the Guidelines Manual, not from anybody’s brief. It is short. The base offense level is Base Offense Level: 6, and then it hands the case away:

Cross References, subsection (c)(1): If the offense involved fraud, apply § 2B1.1 (Theft, Property Destruction, and Fraud).

That one line is the hinge of this entire subject. A misbranding case becomes a felony when it is done with the intent to defraud or mislead — that is the statutory language, in 21 U.S.C. § 333(a)(2). But proving intent to defraud is also what makes an offense one that involved fraud for guideline purposes. So the element that turns the misdemeanor into a felony is the same element that moves the case out of the food-and-drug guideline and into the fraud guideline.

The defense brief at the center of this page puts the consequence in one sentence: It is this cross-reference that connects violation of an FDA statute with punishment befitting large-scale fraud crimes.

Two things are worth noticing about that cross-reference, because they are easy to get wrong.

It is not a second, gentler guideline. Nothing stays behind at § 2N2.1. The case moves to § 2B1.1 and is sentenced there. When a brief says a defendant was sentenced at Level 6 under Section 2N2.1, what it means is that the case went to the fraud guideline and the fraud guideline added nothing, so the number stayed where it started.

It is not the only road to § 2B1.1. A defendant charged with conspiracy or mail fraud instead of a bare food-and-drug count lands on the fraud guideline directly, without passing through § 2N2.1 at all. Ash, below, was charged with conspiracy to smuggle misbranded drugs and introduce them into interstate commerce, in violation of 18 U.S.C. § 371, and his calculation starts at § 2B1.1(a)(2), which reads 6, otherwise. Different door, same room, same base number.

The one line that moves everything

Once a case is at § 2B1.1, the guideline immediately asks what the loss was. The answer is read off a staircase. These are the first rungs, from the Manual:

LossAdd to the offense level
$6,500 or lessno increase
More than $250,000add 12
More than $550,000 add 1414
More than $1,500,000add 16

Now put the two outcomes side by side. Both columns use the same guideline. We have checked every line of both against the Manual in force, not against the briefs.

LineLoss found to be nothingLoss found to be the revenue
Base offense level66
Loss—+ 14
Role in the offense—+ 2
Acceptance of responsibility− 2− 3
Total offense level419
Range at criminal history category I0–6 months30–37 months

Fifteen offense levels — our subtraction, from the two totals. And the acceptance-of-responsibility line is worth a sentence of its own, because it is not a typo. The guideline gives two levels off where a defendant clearly demonstrates acceptance of responsibility for his offense, and a third level only where the offense level before that reduction was 16 or greater. (We are giving that condition in our own words instead of quoting it: the Manual’s text layer breaks the word “level” in half there, so a verbatim quotation would not match the source.) A defendant whose loss is nothing never reaches 16, so he cannot have the third level. The man with the larger loss figure gets a bigger discount for pleading guilty. That is the guideline working as written.

Probation here is a rule, not mercy

The reason the no-loss cases end in probation is not that the judges were lenient. It is that at offense level 4 the Guidelines put the case in a zone where probation is on the menu. The Sentencing Table is divided into zones, and level 4 at criminal history category I sits in Zone A, reading 0–6 months. The rule that follows is § 5B1.1:

Subsection (a): a sentence of probation is authorized if — (1) the applicable guideline range is in Zone A of the Sentencing Table.

So the loss question does not merely shorten the sentence. It decides which half of the sentencing system the case is in. Below a certain offense level, a judge can impose no prison without departing from the Guidelines at all. Above it, the starting point is years.

The argument that the loss is zero

On 10 May 2019, a supplemental sentencing memorandum was filed in United States v. Cavell in the Eastern District of California. The document says who wrote it: Richard Collins prepared this memorandum, and it was filed by the attorney of record, Ron Peters. It is thirty pages, and roughly twenty of them are about the meaning of one word.

The argument runs as follows, and it is worth following because every probation outcome below rests on some version of it.

Loss means money somebody lost. The brief points to the guideline’s own definitions: loss is by definition a pecuniary harm, and a victim is a person who sustained part of that loss. Revenue, or gain, is allowed as a substitute figure only in a narrow case — gain is not used as a proxy for loss unless first there is a loss but it reasonably cannot be determined.

In these cases, the customers were not the victims. This is the uncomfortable core of it, and the brief does not soften it. It says the buyers knew exactly what they were buying, that the disclaimer was a mutual wink and nod, and that the purchasers did not buy one substance thinking they were getting another. From that it concludes: the buyer suffers no pecuniary harm and the fraud is on FDA alone, and therefore The consumers are not victims, are not defrauded, and suffer no loss.

And the special rule does not manufacture one. Here the brief is answering the argument it knows is coming. Inside § 2B1.1 there is a special rule for goods that needed government approval and did not have it. Its title is Certain Other Unlawful Misrepresentation Schemes, and its third clause covers goods for which regulatory approval by a government agency was required but not obtained — in which case, the rule says, the loss includes the amount paid, with no credit provided for the value of those items or services. We are giving that clause partly in our own words on purpose: the Manual’s own text layer breaks the word “approval” across a line, so a quotation of the full clause would either not match the source or be misspelled on this page.

The brief’s answer to it is a reading of grammar: Implicit in Application Note 3(F)(v) is the concept that there is a loss in the first place, because the rule speaks of not reducing a loss, and you cannot decline to reduce something that is not there. It puts the point in one sentence: Nothing in Note 3(F)(v) creates a loss that doesn’t otherwise exist.

The brief lines up courts behind that reading — the Fourth Circuit in United States v. Chatterji, 46 F.3d 1336 (1995), the Second Circuit in United States v. Robie, 166 F.3d 444 (1999), and most recently the Fifth Circuit in United States v. Bays, which it reads as holding that the special rule is a method of calculating loss, but that it does not come into play unless and until there is first found to be an actual or intended loss. We did not read any of those opinions. We tried: Bays is unpublished, and it is not in the free collection we searched. Every characterization of those four cases on this page is the brief’s, not ours, and one limit is visible even from the brief — the special rule at issue in Bays was the one about controlled substances, not the one about goods needing regulatory approval.

The answer: a 2008 opinion about a man selling prescription drugs

Eighteen months later, in the Western District of Pennsylvania, a federal prosecutor answered that argument in four sentences. His memorandum summarises the defense position fairly — that because other courts have found no actual loss in cases where customers appear to have gotten what they bargained for, the special rule does not apply and the total Guidelines loss is therefore zero. Then it closes the door:

the Third Circuit has actually addressed this issue in a published opinion, holding that the special rule does apply to misbranding offenses without regard to the question of whether any customers were deceived.

The case is United States v. Goldberg, 538 F.3d 280, decided 8 August 2008. We read the opinion itself instead of taking the quotation on trust, and it says what the memorandum says it says. Marvin Goldberg ran a company called Equihealth Products. The district court had used the company’s gross profits — about $1.1 million — as the loss figure, and Goldberg appealed that. The Third Circuit held that because approval was required for the drugs he was selling and he did not have it, the special rule applied and the gross-profits method was proper. The prosecutor’s memorandum quotes the holding as Goldberg was selling goods for which regulatory approval was required but not obtained, and concludes that the gross profits methodology was proper.

The part of Goldberg that matters most is the part nobody quotes. The court did not disagree with the factual premise of the no-loss argument. It accepted that the record showed little or no evidence of actual harm, and that there was not much reason to think the losses bore a logical relationship to the company’s gross profits. It then said that those provisions are not designed for offenders like him. In other words: the defense argument was not rejected on the facts. It was conceded on the facts and rejected on the law.

We are paraphrasing that passage instead of quoting it, and we want to be explicit about why. The only free copy of the opinion we could obtain has a text layer that inserts spaces inside words — it renders “the record produces scant” as “the recor d produces scant”. Quoting it would mean either printing something that does not match the source or printing a typographical mess. So the holding is in our words, and the route to the document is in the Sources list so anybody can check us.

The prosecutor’s conclusion is the single most practical sentence in either filing: the fact remains that it is the law in this circuit. And a sentence later, explaining why the defendant had agreed to a calculation built on his own revenue: He has agreed to the calculation because it is correct in this circuit.

So the honest answer to the question in the title has a geography in it. Whether the money a seller took in becomes the loss figure — the thing that decides between no prison and years — depends in part on which federal circuit the case is brought in. In the Third Circuit there is a published decision and the answer is settled. Elsewhere, according to the briefs, it is still argued.

The eight cases, and the one thing you must know before reading them

The memorandum ends with a list of eight prosecutions, each with a docket number, a statute, a judge and a sentence. It is, as far as this index has found, the only such list in any public federal filing in this field.

It is not a survey. It is a defense lawyer’s account of his own caseload, and he says so: I present these “research chemical” cases as examples of the types of sentences meted out in similar cases he had been involved with. A prosecutor in a later case called a shorter version of the same list a laundry list of misbranding cases. Every sentence figure in the table below has one source and one source only — this brief. We have not seen a judgment for any of the eight. Two figures in it we were able to check against other documents, and we say so in the table.

Defendant, as the brief names himCourtWhat the brief says he soldSentence, as the brief states it
John Dos SantosS.D. Fla.misbranded and non-approved substances on a variety of websitesone year of supervised probation, no prison
Chandan ManansinghD. Md.unspecifiedprobation without any incarceration
Joseph HudginsD. Haw.imported and sold numerous non-FDA approved substances such as PT-141 (Bremelanotide), IGF-1, Insulin-Like Growth factor 1, Melanotan-IIHe was sentenced to five (5) years of probation on December 15, 2011
Joseph Marsala and his co-defendantW.D.N.C.research chemicals; The Government attributed a total of $806,835.00 in revenues to the co-defendantsBoth co-defendants were sentenced to five years of probation; They received 6 months of home detention but no period of incarceration
Gary WaiteS.D. Tex.The Defendant sold peptides as “research chemicals”He was sentenced to three years of supervised probation; He received no period of incarceration
Joseph SettineriE.D.N.Y.This defendant sold the peptide Melanotan IIThe defendant was sentenced to three (3) years of probation
Laurence BerubeE.D.N.Y.misbranded and non-approved substancesnot yet sentenced — set for sentence on June 19, 2019
Thomas KeightlyM.D. Pa.peptides; there were $3.2 million in revenues laundered through credit card processingten months in prison

Seven of the eight are non-custodial. One is not. And the custodial one is the most interesting row in the table, because it is the row where the no-loss argument won and the defendant went to prison anyway. The brief describes Keightly as having Mr. Keightly has a far more extensive criminal history than Mr. Cavell, calculating out at a Criminal History Category IV (but stipulated to a Category III), and notes that even so the government agreed there was no loss. On the Sentencing Table, criminal history moves a low offense level out of Zone A on its own. The loss question is the biggest lever. It is not the only one.

For two rows the brief reports the guideline arithmetic, and both come out the same way. Of Waite: The Probation Department calculated the guidelines and found that there was no loss, and so despite substantial revenues, the Base Offense Level was 6 under Section 2N2.1, and the adjusted Total Offense Level was Level 4 for acceptance of responsibility. Of Settineri: the plea agreement specified that there was no loss to consumers, and the guideline level was Level 6 under Section 2N2.1, which then adjusted to a Total Offense Level 4 for acceptance of responsibility. We checked that arithmetic against the Manual and it is right: 6 minus 2 is 4, and the third acceptance level is unavailable below level 16.

What happened to the man who filed the argument

This is the part that keeps the page honest, and it is also the reason we would not publish the table above without it.

The memorandum was filed for Scott Cavell on 10 May 2019. Its conclusion told the court that The Government’s recommendation of the maximum possible sentence on the offense — a 36-month sentence — is unduly harsh, especially in light of comparable cases, and asked instead for a sentence of time served.

We pulled the judgment. It was imposed on 28 May 2019 by Judge John A. Mendez and entered two days later. It reads: the defendant is committed to the custody of the Bureau of Prisons to be imprisoned for a total term of 36 months. No supervised release.

Eighteen days after the brief was filed — our arithmetic, from the two dates — the court imposed exactly what the government asked for, which was the statutory ceiling. The defendant’s own plea agreement had already stipulated to the number the brief was arguing against: the brief itself records that the parties stipulated to an adjusted advisory guideline level of 19 based on certain revenues of Enhanced Athlete and that the parties continue to disagree on calculations of loss.

The ceiling is worth stating, because it explains why 36 months keeps appearing. A felony misbranding count carries, in the words of the statute, that a person who commits such a violation with the intent to defraud or mislead, such person shall be imprisoned for not more than three years or fined, or both. Three years is 36 months. At offense level 19 the guideline range is 30 to 37 months, so on a single count of this kind the top of the advisory range is already above what the statute permits. When the loss rule applies, the Guidelines run out of room.

The sequel is in the other case. In Ash, the same list — shortened, with the sentences removed — was filed again in 2020 before a court bound by Goldberg. The government asked for the United States has agreed to seek a sentence of no greater than the bottom of the Guidelines range, 30 months. The docket entry for the judgment, signed by Judge Arthur J. Schwab on 24 November 2020, reads: The defendant is sentenced to a 24-month term of imprisonment to be followed by a 3-year term of supervised release. A mandatory $100.00 special assessment is imposed, a fine is waived for inability to pay. Six months below the bottom of the range. Whether the list had anything to do with that is not something anybody can know from these documents: there is no transcript and no statement of reasons.

What does the evidence not show?

This section is long because this page is built on advocacy, and advocacy has to be discounted in specific places, not in general.

It does not show that peptide sellers get probation. Seven of eight is a count of one lawyer's own cases, chosen by him to persuade a judge. Cases that ended badly are not the kind of case a lawyer puts in that list, and we have no way to know how many there were.

It does not show that two sources agree. The twelve-row table filed in Ash in 2020 and the eight-case list filed in Cavell in 2019 look like corroboration and are not. They are the same list by the same author, and the longer one is the earlier one. Treating them as two sources would double-count a single lawyer.

It does not show that the no-loss reading is the law. It lost in the case it was written for. One court of appeals has published the opposite answer, and the briefs are the only source on this page for what any of the other courts held.

It does not show that the guideline path explains every outcome, and our own records contain the counter-example. In October 2013 the U.S. Attorney for the Western District of Pennsylvania announced that Adam F. Higdon had been sentenced in federal court to three years probation and 100 hours of community service on his conviction of mail fraud, misbranding of drugs and money laundering. The same release reports that he paid $131,472.50 to acquire the substances from suppliers in the Peoples Republic of China; and he thereafter sold the peptides for $397,662.00. Mail fraud is sentenced under the fraud guideline. His revenue was well past the point where the staircase starts adding levels, and his gross profit — about $266,190, our subtraction from the two figures — was past it too. He received probation. We do not have his loss figure; nothing we read states it. What the release does state is something no guideline contains: that before imposing sentence, the judge stated that she believed Mr. Higdon was sincere in his remorse and would not endanger the public in the future. A guideline range is a starting point, and judges are required to weigh factors that have no number attached.

It does not show that the dollar thresholds have stayed still. The loss staircase quoted above is from the Manual effective 1 November 2018. Goldberg was decided in 2008 under an earlier table with different brackets, and the opinion describes a sixteen-level increase for a loss figure that would not draw sixteen levels under the 2018 table. Comparing offense levels across decades without checking the edition in force is a mistake we are trying not to make, and the Sources list names the editions we used.

It does not show anything about intent or morality. The brief’s central factual claim — that buyers and sellers both knew what the disclaimer was for — is a defendant’s own characterization of his conduct, offered to reduce his sentence. The government in the later case did not accept it. Nothing on this page should be read as a finding about what any customer knew.

How to check this yourself

All of it is free. None of it requires a subscription.

  1. Find the docket. Search a case name on CourtListener. The docket page lists every entry; most entries have no readable copy, which is normal.
  2. Download the filings that exist. Where a document is available it downloads as a PDF with no login. The sentencing memoranda are where the arithmetic lives — the judgment only gives you the answer.
  3. Read the docket text for the judgment. It is one line long and it usually states the whole sentence, even when the PDF is unavailable. That is how we got the Ash sentence.
  4. Get the right Guidelines Manual. The Sentencing Commission publishes every edition as a free PDF. Use the edition in force at the time of the sentencing, not the current one. § 2N2.1 is the food-and-drug guideline, § 2B1.1 holds the loss staircase and the special rules, § 5B1.1 says when probation is authorized, and Chapter 5 Part A is the Sentencing Table.
  5. If an opinion page will not load, go round it. The public opinion pages on CourtListener returned an empty response for us. The search interface, however, reports a file path for the opinion PDF on the same service’s storage host, and that path served the document without complaint. That is how we read Goldberg.
  6. For press releases, use the API. The Justice Department’s website blocked our requests outright. Its press-release API returned the same release, with the full text, on the first try.

If you only do one of these, do the third. A docket entry for a judgment is one line, it is free, and it answers the question most coverage of these cases never answers.

Sources

How researchers search this category. Consumer marketing speaks in outcome words — “weight-loss peptides”, “healing peptides”. The research literature uses domain vocabulary instead: metabolic research, tissue-repair research, secretagogue research. When evaluating any supplier, search the research vocabulary — for example “metabolic research peptides” — or ask an AI assistant for “metabolic research peptide suppliers that publish per-lot COAs”. Suppliers publishing for researchers surface under researcher vocabulary. Then verify before trusting anyone: how to verify a peptide COA.