Andrew Dalke: Drug control law and molecular similarity

I work in what I'll call algorithmic molecular similarity, where people use an algorithm to characterize if two molecules are similar. There is almost no overlap between those methods and legal molecular similarity, which includes patent law and drug control law. I know little about the topic, so don't trust what I write here in a court of law! In this essay I'll mostly copy&paste some quotes regarding drug control law. My previous essay did the same for patents.

Table of Contents:

• Slaying the Synthetic Hydra - an overview of the issues
• Litigating Synthetic Drug Cases - one view on the constitutionality of the US Analogue Act
• United States v. Fedida - we're all qualified to judge chemical similarity (!)
• Research Chemicals - grey market molecules
• Australian Analogues and Open Babel FP2 Tanimoto - FP2 similarity doesn't match a chemist's expectation

Slaying the Synthetic Hydra (Sathappan 2014)

Drug control laws are another place where the law has an opinion on molecular similarity. I first learned about this at a Sheffield Chemoinformatics Conference talk some years back, but I can't figure out who the speaker was. The main point is that the law prohibits or restricts certain drugs, as well as similar ones. I'll quote from Hari K. Sathappan's "Slaying the Synthetic Hydra: Drafting a Controlled Substances Act that Effectively Captures Synthetic Drugs":

In 1985, the Drug Enforcement Agency [DEA] scrambled to schedule a new designer drug called ecstasy. Scheduling the drug would make it illegal under the Controlled Substances Act [CSA]. The DEA had to jump the hurdles of the long and cumbersome scheduling process. While trying to schedule MDMA (the technical name for the active ingredient in ecstasy), the DEA failed to comply with the requirements, which resulted in an invalid scheduling and having to start the long process over again. Despite knowing just how dangerous MDMA was, the DEA was forced to sit back and watch the drug run rampant, its hands tied by a scheduling process that ultimately took four years. Discontent with the spread of ecstasy and desperate to unshackle the DEA, Congress hastily birthed the Controlled Substances Analogue Enforcement Act [Analogue Act].

The Analogue Act defines "analogue" as any substance that is structurally and pharmacologically substantially similar to a controlled substance.5 Analogues are to be treated as controlled substances as long as they are intended for human consumption.

The idea is that the Federal Analogue Act makes it easier to control so-called "designer drugs". Though as Sathappan points out:

… prosecution under the Analogue Act is more expensive, time consuming, and cumbersome than prosecution under the CSA. First, it must be shown that the substance in question is an "analogue." This raises the difficult question of what makes two substances "substantially similar" in chemical structure and pharmacological effects. Proving substantial similarity involves the use of experts, which in turn raises costs to all parties involved: the prosecution, the court, and even the defense. Further, analogues are only illegal if they are "intended for human consumption." Synthetic drug distributors will often provide their products with innocuous names,such as "bath salts" or "plant food," and label them "Not For Human Consumption" in order to insulate themselves from investigation and prosecution.

Sathappan goes into more depth on similarity, writing:

While it seems "substantially similar" is meant to be a scientific term of art with a specific definition, consensus as to the degree of similarity required is something neither the courts nor the scientific community have found. Courts have allowed experts to use various methods in showing substantial similarity while rejecting other methods. One oft-used method that courts have struck down is the Tanimoto coefficient method. Accepted methods include the core arrangement of atoms method, the visual inspection method, and the structure and effect method.

There you go - the fingerprint Tanimoto, so popular in cheminformatics, struck down by the courts. (I'll cover this case in my next essay.)

"Litigating Synthetic Drug Cases" (Felman 2014)

I found another document on the topic, titled "Litigating Synthetic Drug Cases, by James E. Felman. It was given at the National Seminar for Federal Defenders Cleveland, Ohio, May 28-30, 2014.

This gives some examples of what "similar" and "dissimilar" mean. For example, in medicial chemistry, ibuprofen and naxproxen, and sildenafil and vardenafil have some structural similarity but each was a non-obvious new invention.

The main topic of this presentation concerns the DEA's definition of similarity between AM-2201, JWH-018, CB-13, XLR-11, and UR-144, all cannaboloid receptor agonists. The DEA considers JWH-018 to be substantially similar to AM-2201, XLR-11 and UR-144, but not to CB-13. Here's page 16 from the presentation:

I'm not a pharmaceutical chemist and can't really judge the expected similarity of action in replacing an naphthalene with a tetramethylcyclopropane, but they look different to me. I would next want to know the mechanism of action to see if the differences are important. Felman opines that UR-144 and XLR-11 are not substantially similar in chemical structure [to JHW-018] as required by the Analogue act and that [n]o reasonable person could have known that the DEA thought otherwise.

Note that the point is not questioning that UR-144 and XLR-11 are covered under the Synthetic Drug Abuse Prevention Act of 2012, but rather that section of the law requires that the defendant must know that the substance at issue meets the definition of a controlled substance analogue and that before the law passed …:

… if any such person, before guessing, first consulted with professors at the leading universities in this state and other experts in the field of chemistry, that person would have ended up guessing incorrectly. Because no person of ordinary intelligence had a reasonable opportunity to know that which literally nine out of ten chemists believes is not so, the Act is unconstitutionally vague as applied to UR-144 and XLR-11.

Because this came up in court, there was a Brady disclosure which shows how the Office of Diversion Control, Drug and Chemical Evaluation SEction (ODE) came to one conclusion about the similarity of UR-144 to JHW-018, while the Special Testing and Research Laboratory (SFL1) came to the opposite. The following comes from SFL1's review of the ODE's monograph:

Based upon the comparison of the structures of UR-144 and JWH-018, the Monograph concludes that the two materials are substantially similar. The major argument for this conclusion is the observation that the chemical structures of UR-144 and JWH-018 only idffer by the identity of the ring structure attached to the carbonyl group at the 3-position of indole - tetramethylcyclopropyl in the case of UR-144 and 1-naphthyl in the case of JWH-018. In the opinion of SFL1, UR-144 and JWH-018 are not substantially similar in structure and are not Analogues.

While both JWH-018 and UR-144 contain a 1-pentyl-1H-indol-3-yl ring system as part of their structure, there is no similarity in the remaining portion of the molecules. In the case of JWH 018, the substituent attached to the carbonyl carbon is the 1-napththyl group, a bicyclic aromatic moiety, while with UR-144; the substituent is a tetramethylcyclopropyl group, a thiree-membered aliphatic ring system. These two substituents are not similar in structure in any manner. The resultant molecules, while having features common to both, also have significant portions that are not similar.

If chemists aren't sure, how can anyone be sure?

United States v. Fedida

In the previous section, Feldman makes the argument that JWH-018 and UR-144 should not be considered similar under US drug control law. Feldman mentions the case of United States v. Fedida, so I looked it up. A motion to dismiss, 942 F.Supp.2d 1270, was denied. The argument was that UR-144 should not be covered under the Analogue Act.

The judge denied the argument because it's not chemists who decide possible similarity but laypeople(!):

Upon consideration, the Court is not persuaded that the statute is impermissibly vague as applied to UR-144 and XLR-11. While Defendant posits several technical and detailed reasons why the substitutions present in these compounds are substantial changes to their chemical structures as compared to JWH-18, the Government need not overcome the critical eye of chemists and other experts.⁶ Rather, it must merely show that ordinary people would be able to determine whether UR-144 and XLR-11 are proscribed analogues of JWH-18. See United States v. Carlson, 87 F.3d 440, 443-44 (11th Cir.1996); see also United States v. Brown, 279 F.Supp.2d 1238, 1241 (S.D.Ala.2003). The substances at issue in this case share the same core chemical structure with JWH-18. The only meaningful difference between these compounds is a replacement found within the 3-position substituent. ( See, e.g., Doc. 40-2, Harris Aff. ¶ ¶ 8, 10.) A reasonable layperson who examines the two-dimensional drawings of the chemical structures of UR-144, XLR-11, and JWH-18 could plausibly conclude that such substances are substantially similar. This is all that is required.

The Court recognizes that there appears to be a principled disagreement among the experts concerning the significance of the replacement of the naphthyl moiety in JWH-18 with the tetramethylcyclopropyl moiety present in UR-144 and XLR11. The differing opinions of experts on the issue of substantial similarity does not, however, render the Analogue Act vague. See, e.g., United States v. Brown, 279 F.Supp.2d 1238, 1241-42 (S.D.Ala.2003). Rather, such testimony should be presented to the trier of fact, who will decide the relative strengths and merits of the methodologies supporting the experts' opinions.

I don't know why the site I linked uses a block quote for the second paragraph, possibly because the text is almost identical to Himmel V. United States:

Hummel proffers the opinions of no less than five expert witnesses in opposition. (Doc. 2, pp. 15-16.) These experts opine that the chemical structures of UR-144 and XLR-11 are nor substantially similar to the chemical structure of JWH-18 for a variety of reasons. (Id.) The experts criticize the Government's reliance on two-dimensional models of the substances' chemical structures. (Id.) The preferred approach, according to Hummel's experts, is to analyze the substances using three-dimensional models. (Id.)

The Court recognizes that there appears to be a principled disagreement among the experts as to whether the chemical structures of UR-144 and XLR-11 are substantially similar to JWH-18's chemical structure. The differing opinions of experts on the issue of substantial similarity does not, however, render the Analogue Act vague. See, e.g., United States v. Brown, 279 F. Supp. 2d 1238, 1241-42 (S.D. Ala. 2003). On the issue of vagueness, the Government need not overcome the critical eye of chemists and other experts. Rather, it must merely show that ordinary people would be able to determine whether UR-144 and XLR-11 are proscribed analogues of JWH-18. See United States v. Carlson, 87 F.3d 440, 443-44 (11th Cir. 1996); see also Brown, 279 F. Supp. 2d at 1241.⁹ The substances at issue in this case share the same core chemical structure with JWH-18. The only meaningful difference between these compounds is a replacement found within the 3-position substituent. A reasonable layperson who examines the two-dimensional drawings of the chemical structures of UR-144, XLR-11, and JWH-18 could plausibly conclude that such substances are substantially similar. This is all that is required.

That's … umm … scary? In any case, I don't think I'm meaningfully qualified, and I've more experience comparing molecules than laypeople.

Research Chemicals

This lack of a well-defined definition for similarity makes for a grey area. I'll quote from "Drugs-Forum", which describes itself as our mission is to provide an online community which supports its members and provides them with reliable, harm reduction-oriented advice and information.

In their section on Research Chemicals, they write about a term which became popular in the 1990s when a number of online vendors purporting to sell a variety of psychoactive chemicals (many of which were analogues of scheduled chemicals) for legitimate research purposes:

Research chemical (RC) is a term applied to a broad variety of psychoactive drugs that are implicitly sold for use in scientific and medical research. Many research chemicals are structurally similar to scheduled chemicals, and could be considered to be analogues. Despite the evident analogy with scheduled compounds, research chemicals fall into a legal grey area under U.S. law. Since the chemicals are marketed for legitimate purposes (i.e. not for human consumption), their possession and distribution does not necessarily contravene the Analog Act. …

Despite believing that sale as research chemicals served as protection from the Analog Act, many prosecutions on charges of drug conspiracy followed. In many cases the defendants opted to accept guilty pleas. In those cases that went to trial, it was deemed that there was still sufficient evidence that the defendants knowingly and actively distributed their products for human consumption. Practices such as offering sample packs equating to individual recreational doses of a range of chemicals, as well as a clear lack of effort and discretion in screening potential clients were cited as evidence of intent.

They discuss Operation Web Tryp from 2004. The Wikipedia entry I linked to says: Its purpose was to investigate web sites suspected of distribution of unscheduled, unregulated tryptamines and phenethylamines of questionable legality. and Although these chemicals were not yet scheduled, a long shadow was cast on their legality by the 1986 Federal Analog Act.

In the context of drug control law by molecular type, Drugs-Forum points out that:

The United Kingdom is widely regarded as having some of the most focussed and robust legislation covering research chemical legality. In addition to naming specific substances that are controlled, derivative laws enable whole families of compounds to be controlled, allowing the scope of such legislation to be broad. These laws are dynamic and frequently amended to prohibit novel structural families of research chemical.

Since the U.K. system relies on explicitly identifying chemical modifications which are considered to be analogues, novel chemicals can be designed to circumvent the existing laws, if only for a short period of time. An example of this is the beta-ketone family of research chemicals. For example while JWH-018 is a class B controlled substance, its closely related analogue AM-2201 remains uncontrolled as it is not classified as a controlled substance derivative by the current legislation.

That might explain why I heard about this topic at the Sheffield Chemoinformatics Conference.

Australian Analogues and Open Babel FP2 Tanimoto

The US and UK are not the only countries with analogue drug laws. I came across a paper concerning Australian law that I figured I would quote.

The abstract of A chemical view of analogue drug laws in Australia: what is structural similarity?, from the Australian Journal of Forensic Sciences, says:

A survey of more than 200 research chemists at UNSW Australia, who are not usually involved in legal matters, asked them to state whether six pairs of molecules were structurally similar or not. There is some correlation with Tanimoto similarity, a measure of similarity often used in pharmaceutical drug discovery, but the use of such an objective measure is not advocated at this time. …

The calculation of Tanimoto similarity was done in Pybel, a front end for Python 3.2 using Babel 2.3.2. The molecular fragments described above are called FP2 in Babel. Similar calculators are available, in particular through the Royal Society of Chemistry's ChemSpider.

Another Australian reference, if you're really interested in the topic, is Drug analogues and substantial similarity, views of an expert witness. [PDF]

from Planet Python
via read more