Moving beyond immunoassays for poisoned patients

“Urine drug testing is just not adequate to cover what’s going on when we talk about toxic exposures,” Dr. Lynch agreed. A urine drug screen will detect only a small subset—psychostimulants, narcotics, heroin, cocaine, alcohol—of the more than 2.7 million agents responsible for the more than 2 million toxic exposures reported annually to the American Association of Poison Control Centers, she said. “Pharmaceutical events are typically the cause, but we can’t detect most pharmaceutical agents using a urine drug test.”

In a case of an unidentified overdose, an untargeted data acquisition approach to detecting compounds by an LC-HRMS method on a QTOF-MS leads to the most accurate results, Dr. Lynch said.

GC-MS—“the old gold standard”—has its advantages, among them the reproducibility of the mass spectra created for each compound and the large, transferable mass spectral libraries, but “the disadvantages are huge when it comes to the clinical laboratory,” Dr. Lynch said. GC-MS cannot detect nonvolatile, polar, and thermally labile compounds, and “almost all drugs fall into these categories.” Lengthy sample preparation delays results.

Multiple reaction monitoring (MRM) is the most common mass spectrometry method used for compound detection. The drawback is the number of MRM transitions—in which selected precursor ions undergo fragmentation to become product ions—required to detect large numbers of compounds. “You have to do a lot work to set up the method, and it gets bulky with all of these different transitions.”

Another MRM approach uses information dependent acquisition (IDA). With IDA, parameters can be set up to tell the instrument which compounds to fragment, and then all the fragments are collected for a full mass spectrum. “It scans all the product ions, and you get an actual spectrum for that particular compound that can be searched against the library,” Dr. Lynch said. “It’s still a targeted data acquisition approach because you have to tell the instrument what you want to detect.”

With LC-HRMS, there are different types of platforms and options for scanning for compounds. Dr. Lynch and colleagues use an untargeted data acquisition approach on a QTOF, which provides the most information about unknown compounds. The process begins with a full scan, and then the first quadrupole in the instrument acts as a mass filter, selecting a narrow mass range. Users can set the method in different ways, “for example, picking the top 20 ions at any moment in time,” she said. The selected mass is fragmented by collision-induced dissociation in the second quadrupole and sent through a flight tube. “The time it takes for the ion to travel through the flight tube is proportional to the mass-to-change ratios. It gives an accurate measure of the mass of both the precursor ion and the product ions,” resulting in a high mass resolution for more precise detection.

A nominal mass instrument does not help differentiate among compounds with similar mass peaks, Dr. Lynch explained. For example, lamotrigine, diphenhydramine, and 2C-T-4 (2,5-dimethoxy-4-isopropylthiophenethylamine), a designer amine, all have a nominal mass of 255 by LC-MS/MS. “With a high-resolution mass instrument, you get very tight mass peaks and the accurate mass of a compound,” she said. “We would get distinct mass peaks for these three different compounds because they have different formulas and different exact masses.” (See box above.)

An accurate mass helps determine the compound’s formula, narrowing the possibilities. Isotope pattern analysis is also possible. “You have several different ways in which to identify a compound when using this technology.”

LC-HRMS analysis begins with an untargeted data collection. The analysis itself can be performed by one of three methods: targeted, suspect, and untargeted.

For the targeted method, the ZSFG laboratory has a list of about 350 compounds with analytical standards. “We have tested them on our method and know what our limit of detection is and the matrix effects. We’ve validated the assay for those compounds. We have those spectra in our library, so we know everything about it and how well we can detect it.” The process is set up for quick results.

Suspect data analysis is useful when a clinician has an overdose patient and a sample of the drug the patient took but doesn’t recognize the name, which is often the case with new designer drugs. “We can say, ‘We don’t have a standard for that compound. We’ve never seen that compound before,’ and still query the data and see if we see a peak and have acquired spectra for that compound,” Dr. Lynch said. If the drug’s formula is available, it can be used to search the data to see if there is a peak for that formula. If no spectra for that compound are in the library, “we can search the literature and there are other ways to identify or confirm if the compound is there.” Eventually, the laboratory orders an analytical standard to confirm the compound’s presence.

Dr. Lynch said her laboratory tries to steer clear of untargeted data analysis. “It’s like trying to find a needle in a haystack. If you tell the software to pull out the thousand most abundant masses in a particular sample, it will pull out 1,000 masses and give you 1,000 formulas. Then you have to figure out if any of those formulas corresponds to a drug that may have caused the presentation.”

Dr. Lynch is often asked if high-resolution mass spec is as good as tandem mass spec, LC-MS/MS, in terms of its sensitivity. She and colleagues compared the two (Thoren KL, et al. Clin Chem. 2016;62[1]:170–178).

“We took our subset of drugs that we were most interested in detecting, spiked them into drug-free urine in increasing concentrations, and looked for the limit of detection for the two different methods,” she said. Both technologies had the same limit of detection for 76 out of 169 compounds tested. The tandem mass spec method had a lower LOD for 60 compounds, and the LC-HRMS method had a lower LOD for 33 compounds. The differences in LOD between the two technologies were minimal, she said, usually 5 ng/mL or 10 ng/mL.

Regarding specificity, both technologies found 469 of 562 confirmed compounds in 100 urine samples. The LC-HRMS method identified fewer drugs (515) compared with tandem mass spec (596), but LC-HRMS had a 97 percent confirmation rate while tandem mass spec had an 89 percent confirmation rate. “No method is perfect,” she said.

Of the 62 drugs that the LC-HRMS method missed, 33 were unique drugs, and few misses were due to LOD. “For the majority, it was because no LC-MS/MS spectra were acquired. We could see the mass peak, but the instrument failed to acquire the spectra.” She and her colleagues tried different technologies and succeeded in acquiring identification of five additional compounds using a variable SWATH (sequential window acquisition of all theoretical fragment-ion spectra) method (Whitman JD and Lynch KL. Clin Chem. 2019;65[7]:862–870).

Dr. Lynch presented a case from a few years ago in which a 27-year-old male and a 37-year-old female arrived at the Zuckerberg San Francisco General Hospital ED with complaints of extremity weakness and paresthesias. They had fallen asleep after ingesting alcohol, cocaine, and Xanax (alprazolam) with a friend. When they woke up six hours later, their friend was dead. Standard laboratory test results showed similar symptoms in both patients: demand cardiac ischemia, slightly elevated troponin levels (1.43 ng/mL for the female patient, 1.9 ng/mL for the male patient), rhabdomyolysis with elevated CK >1,000 U/L, and compression neuropathy. The male patient also had acute kidney injury (creatinine: 1.68 mg/dL).

Over the next two months, nine additional patient cases and four deaths (the medical examiner investigated the latter) were associated with what turned out to be counterfeit Xanax. Using LC-HRMS untargeted data collection on samples of urine, serum, and pills from eight patients, Dr. Lynch and colleagues determined there was no alprazolam in the samples but detected (along with cocaine and its metabolite and cocaethylene) fentanyl and etizolam. The latter is a designer thienodiazepine and a benzodiazepine analog that produces serious central nervous system depressant effects and is “six times more potent than diazepam and stronger than most of the prescribed benzodiazepines,” Dr. Lynch said. A public health advisory alerted the community.

Overdoses from synthetic opioids other than methadone—nonpharmaceutical fentanyl, illicitly manufactured fentanyl, fentanyl analogs, and novel synthetic opioids—have increased sharply nationwide since 2013. “We have addressed this by ordering a lot of the fentanyl analogs and validating our assay to be able to detect these. So these compounds have been added to our targeted list,” Dr. Lynch said. With a high-resolution platform and an assay, “it’s pretty easy to add to it,” she said. “All you do is order a standard and determine the limit of detection and matrix effects if you’re doing a qualitative assay.” High matrix effects can be expected because of the simple sample preparation required to get quicker results. “It’s a little bit of a tradeoff.”

In an opioid-related case, a 33-year-old male with a history of heroin addiction was found in a bathroom at his worksite. The patient was pulseless but received CPR and naloxone on site and showed slight improvements. The patient later said he had been smoking a drug a friend bought online and admitted to frequently smoking fentanyl. He used the same amount this time because he thought the drug was fentanyl, but his urine drug screen was positive only for amphetamines and THC. LC-HRMS analysis revealed carfentanil in the urine and the drug. Often used to sedate elephants, carfentanil is showing up in adulterated heroin, cocaine, methamphetamine, and illicit Xanax, though it’s a bigger problem on the East Coast and in the Midwest than on the West Coast, Dr. Lynch said, then added, “Our people are just buying and using fentanyl, unfortunately.”

The patient’s blood test results were not positive for carfentanil. “We often request a blood and a urine,” she said. “Sometimes we’ll see something in the urine that we don’t see in the blood, or see something in the blood we don’t see in the urine. When you think about metabolites and windows of detection, the two together are beneficial in helping to determine what’s going on for a patient.” The patient, who was discharged after 24 hours in the ICU, likely had a better outcome because he was not opioid naive.

Another opioid case involved a 29-year-old male found unresponsive after injecting an unknown drug. The patient regained consciousness on the way to the ED and showed symptoms of opiate toxidrome and complained of polydipsia. Standard laboratory results revealed elevated neutrophils and creatinine and a normal eGFR. The urine drug screen was positive for benzodiazepines but not for opiates, despite his presentation.

Audience members thought the mystery drug could be a designer amphetamine, but biological samples tested by LC-HRMS showed positive results for U-47700 (240 ng/mL) and phenazepam (1.4 mg/L). U-47700, also known as Pink, is an opioid seven times more potent than morphine and has been responsible for overdose deaths but often in combination with other drugs, Dr. Lynch said.

“We were curious about the phenazepam because the level was pretty high,” she said. The patient admitted post-discharge to having taken phenazepam, a non-FDA-approved benzodiazepine with a half-life of 49 to 301 hours, a few days before the overdose. He had purchased the phenazepam and U-47700 on the darknet.

While it may be tempting for laboratories to save diagnostic time by ordering several new compound standards for their toxicology libraries, Dr. Lynch cautions against it. “It’s not financially feasible for a clinical laboratory,” she said. Her best advice: “We let the cases lead us.” If she and colleagues get a case and find a compound, they will order the standard and the confirmation.

“There are always new designer drugs, but you may never see a case at your institution, so buying a $500 standard isn’t financially reasonable.” With high resolution, she said, a compound can be identified without a standard. “Follow up with confirmation by ordering that particular standard.”

Amy Carpenter Aquino is CAP TODAY senior editor.