Author: toxlaw

No. Heavy lungs do not independently establish that fentanyl caused a death. In an Orange County series of fentanyl-positive death investigations, combined lung weights were similar whether fentanyl was included in the certified cause of death or not. That comparison also had important selection and confounding limitations.

Which deaths were compared?

Havro and colleagues reviewed casework from 2018–2019, including 285 death investigations with fentanyl detected in central blood. Fentanyl was listed in the cause of death in 238 cases and was not listed in 47. Combined lung weights were available in 275 of the 285 cases.

The lung-weight comparison therefore concerned people already selected for fentanyl detection. It did not compare a representative group of fentanyl-caused deaths with opioid-negative deaths. Availability of lung weights also differed from the complete case count; 275 should not be rewritten as 285 measured lung weights.

The cause-of-death classifications came from the medical examiner’s case determinations. They were not independently assigned through a blinded lung-weight diagnostic experiment.

What were the actual lung weights?

Across available records, combined lung weight ranged from 450 to 3,630 g, with a mean of 1,514 g and median of 1,450 g.

For cases where fentanyl was included in the cause of death, the reported mean was 1,506 g, median 1,460 g and range 560–3,010 g. For cases where fentanyl was not included, the mean was 1,596 g, median 1,400 g and range 450–3,630 g.

These descriptive distributions overlapped substantially. The highest reported lung weight occurred within the group where fentanyl was not included in the cause of death. The authors considered lung weight unreliable for distinguishing the two groups in this dataset. They did not report a validated cutoff, diagnostic sensitivity or specificity for this purpose.

Why the comparison could not isolate fentanyl’s effect

Other opioids were detected in 91 death-investigation cases, including substances such as morphine, methadone, hydrocodone and oxycodone. Those exposures could also be relevant to pulmonary findings. Other diseases and circumstances of death likewise require consideration.

Cases without fentanyl in the cause of death included coronary disease, myocardial infarction, congestive heart failure, hanging and other drug toxicity. A lung weight is consequently not a drug-specific chemical measurement, and this observational comparison did not isolate the effect of fentanyl from every competing explanation.

The authors noted that an opioid-negative comparison group would improve the investigation. Its absence prevents concluding either that fentanyl never affects lung weight or that heavy lungs specifically identify a fatal fentanyl effect.

What did the toxicology establish?

Central blood was collected near the heart cavity at autopsy, preserved with sodium fluoride and potassium oxalate and refrigerated at 4°C. Screening by LC–quadrupole time-of-flight MS was followed by confirmatory quantitative LC–MS/MS. Duplicate quantitative results were averaged.

The fentanyl method used a 0.5–32 ng/mL calibration range, with validated detection and quantification limits of 0.5 ng/mL. Reported bias and precision were within 20% at the tested conditions. These characteristics support interpreting an analytical result; they do not make an anatomical finding specific for fentanyl poisoning.

How to use the finding responsibly

Retain the combined lung weight as one autopsy observation alongside the medical history, full autopsy, treatment history, toxicology and other case evidence. Detection of fentanyl plus heavy lungs is not, by itself, a validated rule for attribution.

A separate issue is whether the fentanyl concentration supplies a fatal threshold. The comparison of living-patient and postmortem fentanyl measurements explains why concentration-only conclusions also need caution.

Primary source

Havro V, Casassa N, Andera K, Mata D. A Two-Year Review of Fentanyl in Driving under the Influence and Postmortem Cases in Orange County, CA, USA. Journal of Analytical Toxicology. 2022;46(8):875–881. doi:10.1093/jat/bkac030.

No. A central-to-peripheral fentanyl ratio near one shows similarity between two postmortem specimens; it does not establish that either concentration equals the concentration during life. An Orange County study illustrates why both the individual range and the collection procedures matter.

What the paired results showed

Havro and colleagues reviewed fentanyl-positive casework from 2018–2019. Their death-investigation dataset included 285 central blood specimens. Peripheral blood was tested in a selected subset of 58 cases, giving 58 central/peripheral pairs.

The central/peripheral ratio had a median of 1.03 and mean of 1.83. Individual ratios ranged from 0.14 to 9.52. The median therefore concealed substantial variation, including cases in which the peripheral result exceeded the central result.

The authors interpreted these ratios as indicating little redistribution in their reviewed material. That conclusion concerns the distribution between sampled postmortem compartments. Without a paired antemortem specimen, a ratio near one cannot demonstrate that neither compartment changed after death.

How were the specimens collected?

Central blood was collected near the central heart cavity at autopsy and refrigerated at 4°C in amber glass bottles. Peripheral specimens were described as coming from a femoral or clavicle artery and were frozen at −20°C in glass tubes. The exact location and mechanism of those draws were not specified.

Blood contained sodium fluoride and potassium oxalate, approximately 2% and 0.75%, respectively, when containers were filled to capacity. These collection and storage differences are part of the comparison. The paper did not provide a uniform death-to-collection interval for the paired subset.

Further peripheral or tissue analysis could be performed when initially detected drugs might have contributed to death and were known to redistribute. The 58 pairs were consequently not a random sample of all deaths or all fentanyl-positive specimens.

What tissue results added

The study also reported 55 brain and 16 liver specimens. Brain fentanyl ranged from 1.9 to 441 ng/g, while liver ranged from 11.6 to 226 ng/g. Those tissue units must remain distinct from blood concentrations in ng/mL.

Fifteen liver/peripheral pairs had a mean ratio of 6.78, median of 6.20 and range of 1.90–11.7. Dividing tissue ng/g by blood ng/mL produces mL/g, numerically equivalent to L/kg. It is not a dimensionless blood-to-blood ratio. These selected paired data do not establish a universal tissue-to-blood conversion or quantify a within-person change from life to death.

Screening and confirmation were separate

Initial screening used LC–quadrupole time-of-flight MS. Fentanyl was subsequently confirmed and quantified by LC–MS/MS, with duplicate measurements averaged for reporting. The quantitative calibration range was 0.5–32 ng/mL. The authors reported validated detection and quantification limits of 0.5 ng/mL and bias and precision within 20% at the tested validation conditions.

Those analytical findings support the measurement procedure. They do not establish biological equivalence between collection sites or identify the cause of an individual ratio difference.

What to retain in a case review

Keep both concentrations, both sites, preservation, handling and any sampling limitations. Review the full paired distribution rather than turning its median into a correction factor. A high or low ratio alone does not establish dose, impairment or cause of death.

For additional collection-site evidence, see subclavian versus femoral postmortem blood. That separate research also requires its own specimen and analyte context.

Primary source

Havro V, Casassa N, Andera K, Mata D. A Two-Year Review of Fentanyl in Driving under the Influence and Postmortem Cases in Orange County, CA, USA. Journal of Analytical Toxicology. 2022;46(8):875–881. doi:10.1093/jat/bkac030.

An 18% figure in a chiral hair-analysis abstract can be misread if the reader omits which compounds had to be present. In Zhang and colleagues’ 2018 study, 18% referred to samples containing four analytes: both methamphetamine enantiomers and both amphetamine enantiomers. It did not describe the entire group with both methamphetamine enantiomers.

Which substances were separated?

The analytical study distinguished S-(+)-methamphetamine, R-(−)-methamphetamine, S-(+)-amphetamine and R-(−)-amphetamine. Methamphetamine and amphetamine are separate analytes, each with two mirror-image forms. A result identifying one cannot silently be substituted for the other, and a total methamphetamine measurement does not reveal its S/R composition.

The researchers developed an HPLC–tandem-mass-spectrometry method and applied it to a selected set of 50 positive hair specimens described in the paper’s results. This was laboratory method research with case specimens, not a controlled drug-administration experiment or a representative population survey.

How do the denominators fit together?

The full-text results describe 35 specimens, or 70% of 50, with only the S forms of methamphetamine and amphetamine detected. The remaining 15 contained both S and R methamphetamine, together with S amphetamine. Nine of those 15 also had R amphetamine detected.

Different detection patterns within the 50 positive hair specimens described in the results.

Pattern	Count	Denominator and percentage
S methamphetamine and S amphetamine only	35	35/50 = 70%
Both methamphetamine enantiomers	15	15/50 = 30%
All four analytes	9	9/50 = 18%
R amphetamine among the dual-methamphetamine subgroup	9	9/15 = 60%

Thus 18%, 30% and 60% can each be correct while describing different questions. The table is an original count-based explanation of the reported findings, not a reproduction of the paper’s chromatograms or specimen-by-specimen table.

Does “not detected” mean an enantiomer was absent?

Not necessarily. The method reported detection limits of 0.1 ng/mg for methamphetamine enantiomers and 0.15 ng/mg for amphetamine enantiomers. Reported quantification limits were 0.4 and 0.5 ng/mg, respectively. These units express analyte mass per mass of hair, not a blood concentration or an administered dose.

The specimen table marks some R-amphetamine results as not detected even when R-methamphetamine was measurable. That pattern does not establish that the precursor drug lacked the corresponding enantiomer. Concentrations below the method’s detection capability and losses during processing remain relevant possibilities. Nor does reporting only S forms establish chemically perfect S purity.

Why is a quantification limit different from a calibration range?

The paper separately reports a linear calibration range of 15–300 ng/mg and lower signal-based detection and quantification limits. It calculated the latter using signal-to-noise criteria of at least three for detection and ten for quantification. Quality-control precision assessments used 20, 100 and 250 ng/mg.

Those are distinct pieces of validation information. A low signal-based quantification limit does not, by itself, demonstrate the same quantitative performance at every concentration between that limit and the lowest reported calibration point. For a case result in that interval, the laboratory’s actual validated calibration and reporting procedures would need examination rather than assuming the abstract supplies the missing validation.

How did preparation affect the measurement?

The investigators compared acid hydrolysis, alkaline hydrolysis and heated methanol ultrasonication. Their chosen procedure used approximately 20 mg of prepared hair, methanol and a 70°C ultrasonic water bath for 30 minutes, followed by chiral chromatography. Methamphetamine peak behavior varied with the extraction method.

This matters because enantiomer interpretation depends on successful preparation as well as separation and mass-spectrometric identification. The study’s favorable extraction comparison is specific to the conditions tested; it is not proof that every laboratory’s hair procedure will recover and quantify all four compounds identically.

What can the case percentages establish?

They show that the method resolved different analyte patterns in these selected specimens. The study did not administer known doses or establish each donor’s formulation, route, timing or product purity. The percentages cannot identify the prevalence of a particular product in the general population or prove an individual’s drug source from the pattern alone.

For a client explanation, identify every analyte required for a quoted percentage, retain its denominator and distinguish detection from quantification. A chiral hair finding is not a contemporaneous impairment measurement, and this method study provides no validated conversion from a hair result to driving ability or a whole-blood concentration.

Scientific source

Zhang Y, Li L, Xing X, Zhou Z, Ma A. Analysis of methamphetamine and amphetamine chiral enantiomers in hair by high performance liquid chromatography–tandem mass spectrometry. Chinese Journal of Chromatography. 2018;36(12):1290–1296. Chinese-language full text with English abstract. doi:10.3724/SP.J.1123.2018.08040. Historical analytical method study; original commentary based on the supplied paper.

A pilot trial can succeed at its intended decision without proving that treatment outperforms placebo. In a 2016 methamphetamine-use-disorder study, “success” meant that a combination of naltrexone and bupropion showed enough promise to justify further investigation under a prespecified two-stage design. Understanding that purpose prevents its response percentage from being mistaken for a definitive treatment effect.

Who received which intervention?

Mooney and colleagues enrolled 49 adults across three US sites. Participants met criteria for severe methamphetamine-type stimulant use disorder and reported methamphetamine use on at least 20 of the preceding 30 days. Screening also required repeated methamphetamine-positive urine results. This was a selected, treatment-seeking population, rather than a sample of everyone with a positive toxicology test.

The intervention lasted eight weeks. Extended-release naltrexone, 380 mg by injection, was administered at weeks 1 and 5. Oral extended-release bupropion increased from 150 mg/day on days 1–2 to 300 mg/day on days 3–4 and 450 mg/day from day 5, with a reduction to 300 mg permitted for tolerability. A taper followed during week 9. These are historical research procedures, not prescribing instructions.

Were the two stages different treatment groups?

No. Stage 1 enrolled 20 participants; stage 2 added 29. Both stages used the same intervention and procedures. Stage 2 would proceed only if at least three stage-1 participants met the response criterion. If fewer did, the study would stop for insufficient promise.

Five stage-1 participants responded, allowing recruitment to continue. Six more responded in stage 2. Thus 11 of 49 met the endpoint, exceeding the final threshold of nine responders. The stages were sequential recruitment decisions, not randomized active and placebo arms, and not a randomized comparison of different treatment durations.

What counted as a response?

A responder needed at least six methamphetamine-negative urine drug screens among eight scheduled during weeks 5–8, including a final negative screen obtained in week 8. Central-laboratory results used a 500 ng/mL cutoff. Study staff did not receive the treatment-phase urine results, reducing one potential source of assessment influence.

This was a defined pattern of urine results during a specified four-week evaluation period. It did not require every sample to be negative, demonstrate uninterrupted abstinence between collections, or establish recovery after medication ended. A negative classification under that cutoff also should not be equated with an analytically established concentration of zero.

Why does the paper report 24% when 11 divided by 49 is about 22%?

The raw proportion is approximately 22.4%. The authors separately reported a design-adjusted unbiased estimate of 24%, calculated using methods appropriate to the two-stage procedure. They reported a 95% lower confidence limit of 13% and a trial p-value of 0.0075.

Those numbers should retain their labels. The lower confidence limit is not a complete two-sided interval. The p-value addresses the design’s hypothesized response benchmark, which assumed a null response rate of 10% or less. It is not a p-value from a contemporaneous randomized comparison with placebo: this pilot had no placebo group.

What else accompanied the medication?

Participants attended clinic twice weekly for observed oral dosing, urine collection, assessments and medical management. Smartphone videos documented take-home dosing, and compensation included payments for qualifying adherence videos. Blood measurements of bupropion and hydroxybupropion provided additional adherence information.

Those procedures helped establish feasibility and medication exposure. They also mean that the observed outcome arose within an intensive research setting. The pilot cannot isolate the contribution of medication from expectations, contact, adherence support or participant selection. It likewise cannot separate the independent contributions of naltrexone and bupropion because everyone received the combination.

Did meeting the research threshold establish safety?

No. Forty-five participants reported 249 treatment-emergent adverse events, and two participants had serious events: convulsion and acute amphetamine intoxication. The authors attributed one generalized seizure to bupropion. It occurred four days after the participant reached the 450 mg maintenance dose; the discussion also identified other possible factors affecting seizure susceptibility.

A small pilot can reveal important adverse events while remaining unable to estimate uncommon risks precisely. Its positive decision to continue development is therefore not a declaration that treatment is risk-free.

How should the result be explained to a client?

The defensible historical conclusion is that the regimen crossed a predefined signal threshold and warranted a more definitive test. Its result is not a cure rate, an individualized prognosis or a placebo-adjusted benefit estimate. Comparisons with later trials must preserve differences in design, dosing schedules, population and response rules instead of comparing percentages as though they measured the same outcome.

Scientific source

Mooney LJ, Hillhouse MP, Thomas C, et al. Utilizing a two-stage design to investigate the safety and potential efficacy of monthly naltrexone plus once-daily bupropion as a treatment for methamphetamine use disorder. Journal of Addiction Medicine. 2016;10(4):236–243. doi:10.1097/ADM.0000000000000218. Historical open-label human pilot study; original research-based commentary.

Bone marrow can contain measurable methanol when blood is unavailable, but a rabbit-derived average ratio is not a validated universal conversion for human deaths. Winek and Susa found that accounting for marrow lipid improved the blood–marrow relationship in 36 rabbits. Individual prediction differences and storage changes remained important.

How the experiment was designed

The investigators gave 36 New Zealand albino rabbits intraperitoneal methanol at 0.75, 1.0 or 1.5 g/kg. Blood was sampled from the marginal ear vein before dosing and at 5, 10, 30 and 60 minutes. The animals were killed 60 minutes after administration.

Separate groups of twelve underwent sampling immediately, four hours or 24 hours after death. At that examination, heart blood and marrow from the hind-leg tibia and fibula were collected. This design compared postmortem heart blood with marrow; it did not use a standard human femoral-blood comparator.

Intraperitoneal dosing also excluded the stomach contents as a source of orally administered methanol in this experiment. That feature limits transfer to human deaths involving oral ingestion, as do the species, exposure route and short survival period.

What lipid correction changed

The overall heart-blood/observed-marrow ratio was approximately 2.6 ± 0.6, with a detailed reported range of 1.49–4.21. Accounting for each specimen’s lipid content reduced the ratio to approximately 1.6 ± 0.3, with a range of 1.16–2.92.

The authors reported that the observed and lipid-corrected ratios differed significantly at the 99.9% confidence level. They did not find a significant difference between the four-hour and 24-hour delayed-sampling groups. That null result did not establish stability over longer postmortem intervals or in decomposed human remains.

Correlation improved after lipid correction, but correlation did not mean that one multiplier reproduced every blood result. The authors reported an average prediction difference of 17.7% using uncorrected marrow and 13.2% using corrected marrow. Their individual results still contained sizable deviations: Table 4 includes a 45% difference in one lipid-corrected estimate.

The publication’s prose summarizes a narrower corrected difference range than that table example. Its data therefore should not be presented as establishing a guaranteed maximum error. More importantly, the estimates were developed and evaluated in the same animal dataset, rather than tested in an independent human validation cohort.

How methanol was measured

The laboratory used direct-injection gas chromatography with flame-ionization detection. Samples were prepared in duplicate and each preparation analyzed in triplicate; the average of six measurements was used. Repeated injections improve assessment of a prepared sample, but do not turn one animal into six independent biological observations.

In eight aqueous controls, sonication produced a mean methanol loss of 2.79 ± 1.65%, ranging from zero to 4.5%. That preparation check was not a complete modern matrix-validation study. The article did not provide numerical detection and quantification limits for this workflow.

Storage altered the specimens

Marrow portions received 3% sodium fluoride and were frozen at −6°C, then thawed and reanalyzed after 24 hours and weekly for three weeks. Methanol changed despite freezing and preservative, usually decreasing. The authors warned that delayed analysis undermined concentration estimation.

This storage observation is specific to the tested conditions. It does not demonstrate equivalent behavior at −20°C, with another preservative or in every marrow specimen.

What human evidence was available?

The paper added one human methanol death encountered during the investigation, with heart-blood/observed-marrow and corrected ratios of 2.2 and 1.9. One compatible observation is insufficient to validate a general conversion across people, marrow collection sites or decomposition conditions.

Marrow detection may contribute evidence of exposure. It does not, by this experiment alone, establish a person’s earlier blood concentration, impairment or cause of death. For a different tissue-ratio problem, see the repeated hydrocodone liver-to-blood outlier.

Primary source

Winek CL, Susa D. Blood versus bone marrow methanol concentrations in rabbits. Forensic Science International. 1982;19(2):165–175. doi:10.1016/0379-0738(82)90042-1.

No. The 57.9 µg/L fentanyl result in a published fatality does not establish a universal lethal threshold. The authors attributed the death primarily to fentanyl using the measured result together with extensive scene evidence and other toxicological findings. One selected case cannot define the concentration that would prove poisoning in another person.

What was observed, and what remained suspected?

Peeters and colleagues described a 23-year-old man found dead with 67 non-prescribed fentanyl patches on his body. Empty medication blisters and powder raised suspicion of additional oral fentanyl and cocaine use. Possible tablet ingestion was an inference from circumstances, not a measured absorbed dose.

The authors estimated the combined nominal patch delivery rate at approximately 1,310 µg/hour. That label-based rate is not the amount demonstrated to have entered the bloodstream. The estimated time between death and examination was approximately 12 hours.

A forensic physician examined the body, but no autopsy was performed. Urine was obtained by puncture. Blood was collected from the subclavian site because femoral blood could not be sampled. This collection-site limitation matters when comparing the result with reports using femoral blood.

Screening, semiquantitation and targeted quantitation

The investigation combined scene urine testing, a laboratory urine drug-of-abuse panel and blood screening using UPLC–mass spectrometry. The blood-screening procedures were described as using plasma. Fentanyl and other substances were identified, followed by a specific validated UPLC–MS/MS assay that produced the reported fentanyl concentration of 57.9 µg/L.

The quantitative assay referenced a plasma method. The number should therefore travel with the subclavian collection site and described plasma processing, rather than being silently relabeled as a femoral whole-blood result. The case report did not reproduce numerical validation limits, precision or measurement uncertainty for this targeted assay.

Benzoylecgonine was approximately 1 mg/L on semiquantitative assessment. Cocaine and ecgonine methyl ester were not quantitatively measured but were reported below 1 mg/L. Those statements do not supply exact cocaine concentrations or establish its independent contribution to death.

Why the ethanol result also needed qualification

An assay described as an ethanol immunoassay gave 0.22 g/L. The authors discussed possible interference and postmortem effects. Headspace GC–FID confirmation was not performed because they considered the result low and less important to the case interpretation.

Accordingly, this was not a chromatographically confirmed ethanol concentration. It should not be used to establish drinking, quantify an alcohol contribution or claim a demonstrated fentanyl–ethanol interaction in this individual. The authors stated that the effects of cocaine and ethanol on the fentanyl concentration remained unknown.

What the authors concluded

The combination of numerous observed patches, targeted fentanyl measurement and circumstances supported the authors’ attribution of fentanyl as the principal cause of death. They retained cocaine as a possible contributor. That is a case-level interpretation, not a finding that fentanyl detection alone proves causation.

The report compared its result with other published fatalities, but the cases differed in sampling site, interval, formulation, suspected co-exposures and clinical history. Such a selected literature comparison does not establish sensitivity or specificity for a threshold of 50 or 57.9 µg/L.

There was no antemortem sample, no serial postmortem sampling and no measured individual redistribution factor. The result cannot reconstruct the earlier circulating concentration, exact absorbed dose or impairment before death.

What to preserve in another case

Record the blood collection site, processed matrix, timing, method and distinction between observed exposure evidence and suspected ingestion. For the separate collection-site evidence, see subclavian versus femoral postmortem blood. A general site difference should not be used as an automatic fentanyl correction factor.

Primary source

Peeters LEJ, Vleut IT, Tan GE, Croes EA, Bethlehem C. Case report on postmortem fentanyl measurement after overdose with more than 67 fentanyl patches. Forensic Toxicology. 2022;40:199–203. doi:10.1007/s11419-021-00598-3.

No. A group comparison showing substantially higher postmortem fentanyl concentrations does not provide a factor for reconstructing an individual’s concentration during life. Andresen and colleagues compared separate postmortem and living-patient groups receiving transdermal fentanyl. They did not measure paired before-and-after concentrations in the same people.

Which people and specimens were compared?

The publication describes a postmortem comparison of 118 cases and serum from 27 living pain-treatment patients. Analyses used smaller subsets because of undetectable concentrations, defective patches or other exclusions. The postmortem group was predominantly elderly and medically ill; 63% had fentanyl prescribed for oncological pain.

Deaths were initially considered unrelated to fentanyl after medical-history review and examinations by two physicians. No routine autopsy was performed in that comparison cohort. One case was retrospectively excluded after suspicion of fentanyl intoxication arose. The authors acknowledged that a contributory role of fentanyl could not be completely excluded in the remaining cases.

Postmortem whole blood came from the femoral vein after dissection and visual localization. Samples were collected one to 41 days after death and stored without preservative at −20°C. Living-patient serum was collected at least 24 hours after application of a new patch, following at least one week of therapy, and stored at 4–8°C. All study specimens were analyzed within three months.

How reliable were the measurements?

Fentanyl was quantified by liquid chromatography–mass spectrometry using fentanyl-d5 as internal standard. The detection limit was 0.9 µg/L and quantification limit 1.0 µg/L. Initial calibration covered 1–15 µg/L, with an extended curve for higher results. Inter- and intra-day imprecision was below 15%; recovery was 84.6% at 2.5 µg/L and 93.7% at 10 µg/L.

The reported concentrations are in micrograms per liter, not milligrams per liter. Quality controls showed no loss outside the study’s accepted variation during three-month temperature comparisons. That control experiment does not independently establish stability under every authentic specimen’s handling history.

What “up to nine times higher” meant

The authors described dose-dependent group-average differences of approximately four- to ninefold between postmortem blood and living-person serum. These were comparisons across different people, specimen types, illnesses and collection times. They were not observed within-person increases caused solely by death.

The 102 measurable postmortem cases with intact patches used in one regression showed a dose–concentration correlation of r = 0.596. Restricting the analysis to doses up to 150 µg/hour reduced it to r = 0.286 in 98 cases, although that association remained statistically significant. In the living group, the dose regression included only 12 of 27 patients with quantifiable, eligible results and gave r = 0.742.

Those findings do not support predicting an individual’s concentration from patch strength. Nor does statistical significance make the smaller postmortem association a useful reconstruction equation.

What the study could not separate

Postmortem redistribution was a plausible explanation for higher concentrations, but the study did not isolate it from differences between the patient groups. Whole blood and serum were also different matrices, even though the authors discussed literature suggesting similar fentanyl partitioning during life.

Physical-build categories did not show a significant concentration difference at the two patch doses examined. Longer postmortem intervals did not inevitably produce higher concentrations. Because the earliest collection was one day after death, the study could not directly investigate the first postmortem hours.

The authors found substantial overlap between concentrations associated with therapeutic administration and reported intoxications. A measured fentanyl concentration alone therefore did not establish impairment or cause of death. Medical history, tolerance, co-medication, patch history and specimen details remained necessary.

Compare the separate problem of hospital versus autopsy cannabinoid measurements. A group comparison and a within-person comparison support different inferences.

Primary source

Andresen H, Gullans A, Veselinovic M, et al. Fentanyl: Toxic or Therapeutic? Postmortem and Antemortem Blood Concentrations After Transdermal Fentanyl Application. Journal of Analytical Toxicology. 2012;36(3):182–194. doi:10.1093/jat/bks005.

Pericardial fluid can provide preliminary drug-screening information when urine is unavailable, but it is not automatically interchangeable with urine. A 66-autopsy study documented benzodiazepine false-negatives in both matrices, including samples containing substantial parent-drug or metabolite concentrations. An automated positive-or-negative printout remained a presumptive result.

What the comparison included

Tominaga, Michiue and Maeda compared Triage-TOX, Triage-DOA and Monitect-9 using 51 urine specimens and 19 pericardial-fluid specimens from 66 forensic autopsies. Seventy specimens did not represent seventy independent deaths. Cases were selected from January 2011 through August 2014 to include a range of instrumentally detected drugs.

The bodies had no evident putrefactive changes. Samples were stored at −20°C until use. The investigators applied manufacturers’ urine-testing procedures to pericardial fluid as well. That experimental application should not be mistaken for proof that urine thresholds have identical diagnostic meaning in another matrix.

The selection included 24 urine and 13 pericardial-fluid specimens containing benzodiazepines or related compounds by instrumental analysis. These totals are different from the smaller subsets used in threshold-based comparisons. Parent drugs, metabolites and assay cross-reactivity all affected whether a screening device responded.

Automation did not eliminate false-negatives

Triage-TOX used an automated fluorescence-immunoassay reader to produce objective results without relying on a person’s interpretation of a faint visual signal. That improved objectivity did not guarantee detection.

The paper describes four benzodiazepine false-negative urine results with Triage-TOX, including specimens containing temazepam at 0.687 µg/mL, flunitrazepam at 0.738 µg/mL, lorazepam at 1.07 µg/mL and nordiazepam at 9.079 µg/mL. In pericardial fluid, reported Triage-TOX false-negative examples contained etizolam at 1.239 µg/mL and nordiazepam at 1.399 µg/mL.

These are specimen-specific drug or metabolite concentrations measured instrumentally. They are not interchangeable with an immunoassay’s class-calibrator concentration, and they do not identify a blood concentration or degree of sedation.

Why a second screen still was not confirmation

Monitect-9 detected benzodiazepines more readily in this selected material, but also produced more results the authors classified as false-positives. Some apparent disagreements required LC–MS/MS because low concentrations or metabolites were not detected by GC–MS. The comparison therefore depended on the reference procedure’s capabilities as well as the screening device.

The authors proposed that combining devices could help reduce misinterpretation. Agreement between two immunoassays, however, is not the same as analyte-specific instrumental identification and quantitation. Both screens can be influenced by their target coverage and cross-reactivity.

Instrumental testing used GC–MS and LC–MS/MS. The reported GC–MS detection thresholds for benzodiazepines and metabolites were 0.01–0.05 µg/mL, compared with 0.001–0.005 µg/mL for LC–MS/MS. The lower thresholds helped investigate discordant findings; they were not universal exposure or impairment cutoffs.

What the study did not validate

The small, deliberately selected specimen set does not establish predictive values for all autopsies. In particular, it did not validate performance in advanced decomposition. The authors also described problems with turbid deteriorating specimens, but those observations did not constitute a systematic decomposition experiment.

Several drug classes had no instrumentally positive specimens, so negative results for those classes could not demonstrate sensitivity. Nor did the study test whether a screening result alone correctly identified the cause of death. The device supplier partly supported the investigation, as disclosed by the authors.

Reviewers should retain the exact matrix, device, target compound, metabolite findings and follow-up method. For a related limitation of instrumental reporting, see why an unquantifiable result is not a negative test.

Primary source

Tominaga M, Michiue T, Maeda H. Evaluation of the on-site immunoassay drug-screening device Triage-TOX in routine forensic autopsy. Legal Medicine. 2015;17(6):499–502. doi:10.1016/j.legalmed.2015.10.007.

No. A negative preliminary autopsy drug screen does not exclude exposure to medication, and a negative result from one instrumental workflow does not exclude every drug. A retrospective study of 2,996 forensic autopsies found additional drugs through broader GC–MS screening and selected LC–MS/MS reanalysis. The findings concern the scope and performance of particular procedures, not a universal failure rate for toxicology.

What did the study review?

Tominaga and colleagues reviewed routine work at an Osaka forensic institute from January 1996 through June 2014. Cases without adequate specimens because of advanced decomposition or skeletonization were excluded. This was an institutional autopsy series, not a random sample of deaths in the general population.

On-site immunoassays and GC–MS were used in the routine program. In selected cases during the final four years, another laboratory performed additional GC–MS and LC–MS/MS testing. Reanalysis was prompted by discrepancies or continued suspicion of drug exposure despite negative screening, when suitable specimens remained available.

Blood and several additional matrices were routinely collected, including peripheral blood, urine, pericardial fluid, cerebrospinal fluid, bone marrow aspirate, bile and stomach contents. Alternative effusions were used when necessary. Specimens were stored at −20°C after preliminary immunoassay testing. The paper does not supply one uniform collection site or death-to-sampling interval for every reported finding.

What broader testing added

GC–MS detected drugs in 486 cases, or 16.2% of the series. The investigators reported 2,440 cases with negative preliminary immunoassays and identified additional drugs by GC–MS in 218 cases—7.3% of all 2,996 cases. These included medications outside, or inadequately represented by, the preliminary panels, such as antidepressants, acetaminophen and salicylic acid.

The denominator matters. The reported 7.3% is a proportion of the complete autopsy series; it should not be described as the sensitivity of an immunoassay for every drug. A medication outside a panel is a different problem from a panel target present above its threshold but missed.

Selected LC–MS/MS reanalysis also showed that the original GC–MS workflow was not an exhaustive reference. In 25 of 32 cases selected for discordant benzodiazepine findings, LC–MS/MS detected low concentrations of benzodiazepines or their metabolites. Seven had no such detection. The detailed results refine the abstract’s broader description of the 32-case reanalysis.

Why those 32 cases do not measure routine sensitivity

These cases were selected because the results or circumstances already raised concern. They were not a random validation sample of all negative GC–MS specimens. Consequently, 25 of 32 cannot be applied as an expected missed-detection rate in another laboratory.

The authors reported GC–MS quantitative precision below 10% for the drugs and specimens evaluated. That describes repeatability within the studied procedure; it does not establish detection of every possible substance. Some LC–MS/MS findings were explicitly semiquantitative and should not be rewritten as definitive quantitative blood concentrations.

The analytical instruments and screening arrangements also changed across the 18.5-year period. Apparent changes in detection over time therefore require attention to ascertainment and methodology, as well as possible changes in exposure.

Detection did not automatically establish poisoning

Of the 486 GC–MS drug-positive cases, 123 were classified as fatal intoxications. Drugs were also detected in deaths from other causes. Drug-category counts overlapped because more than one drug could be present. Neither a positive result nor a count of co-detected medications independently establishes impairment or a causal contribution.

A case review should identify the panel, analytes, matrices, reporting thresholds, extraction and instrumental methods, and reasons for any additional testing. For another specimen-specific screening question, see vitreous humor versus blood for postmortem drug screening.

Primary source

Tominaga M, Michiue T, Inamori-Kawamoto O, et al. Efficacy of drug screening in forensic autopsy: Retrospective investigation of routine toxicological findings. Legal Medicine. 2015;17(3):172–176. doi:10.1016/j.legalmed.2015.01.001.

A molecular association with methamphetamine use is not automatically a test that can identify exposure, diagnose dependence or prove brain injury in a particular person. A 2021 review by Jayanthi, McCoy and Cadet describes an extensive epigenetic research field, but its animal experiments, cell studies and human comparisons answer different questions.

What does “epigenetic” mean in this evidence?

The review examined regulation of gene activity through histone modifications, DNA methylation and hydroxymethylation, and non-coding RNAs. These measurements concern regulation around genetic information. They should not be described as proof that methamphetamine changed the underlying DNA sequence or produced an inherited mutation.

The source was a historical narrative review, not a new diagnostic validation trial. It brought together experimental animal and cell evidence with observational human findings. A study manipulating a molecular pathway in a rat brain can help investigate mechanism within that model. A difference between human patient and control blood samples provides a different kind of evidence and cannot inherit the animal experiment’s causal control.

Why must the tissue and assay remain attached to a result?

The review describes measurements from brain regions including the prefrontal cortex, dorsal striatum and nucleus accumbens, as well as human peripheral samples. Regional effects differed. Some histone-regulating enzymes increased in one brain region while others decreased elsewhere. A single phrase such as “increased epigenetic activity” would discard the biological specificity.

Assays also measured different things: total protein abundance, enrichment at a particular gene promoter, methylation at selected DNA sites, or RNA expression. The review discusses experiments in which promoter-specific findings differed despite no detected difference in broader histone protein expression. A global measurement and a locus-specific measurement are not interchangeable endpoints.

Correspondingly, a peripheral blood observation cannot be relabeled a direct measurement of the same participant’s brain tissue. The review treats peripheral findings as potential windows into biology, while calling for more work to establish their meaning.

What did the human RNA examples show?

One exploratory human study summarized in the review reported 19 increased and 69 decreased microRNAs in plasma extracellular-vesicle research involving female patients. Those figures count RNA features, not patients, diagnoses or independent replications. Another serum study involving male and female participants reported five increased and nine decreased microRNAs. Serum, plasma and an isolated extracellular-vesicle fraction remain different sample descriptions.

Associations with clinical history and symptoms were also not uniform. Some RNA measures related differently to age at first use, lifetime use and recent frequency. A separate plasma-vesicle study related selected microRNAs to anxiety and depression scores. The review specifically called for larger comparisons with other clinical populations who had not been exposed to psychostimulants before assigning those candidate measures a clinical biomarker role.

Why do subgroup and null findings matter?

The human DNA-methylation literature summarized in the review included a parvalbumin finding in the psychotic subgroup of methamphetamine-dependent patients, with no change reported in the nonpsychotic subgroup. That pattern cannot be generalized into a universal marker for everyone who has used methamphetamine.

A candidate marker might reflect exposure, a psychiatric condition, another exposure, a background difference, or several influences together. An association alone does not resolve those alternatives. Likewise, the absence of a difference on a selected marker does not establish that every biological effect is absent.

What would have to change before this became an individual test?

As an interpretation of this evidence, individual use would require a clearly defined target claim and separate validation. Identifying a history of exposure, diagnosing a use disorder, assessing psychiatric symptoms and demonstrating structural injury are different targets. A model that distinguishes selected patients from healthy controls may not distinguish patients from other clinically similar groups.

A usable decision rule would also need a specified specimen, laboratory method, cutoff, sensitivity, specificity and independent assessment in the intended population. The number of altered molecular features does not supply those performance measures. Nor does a statistical difference between group averages tell a reader how much the groups overlap or whether a particular result is correctly classified.

Does a molecular change establish a treatment?

No clinical treatment conclusion follows merely from identifying a pathway. The review discusses genetic manipulations and pharmacological interventions in preclinical systems as possible leads. Correcting a laboratory marker, changing animal activity or modifying conditioned reward does not independently establish improved human function, acceptable clinical safety or durable recovery.

For a client explanation, state what was sampled, what was measured and which comparison produced the result. Label animal mechanisms, exploratory human associations and proposed biomarkers separately. This historical review supports further mechanistic and validation research; it does not provide an individual forensic exposure test, an impairment threshold or proof that a person’s symptoms were caused by methamphetamine.

Scientific source

Jayanthi S, McCoy MT, Cadet JL. Epigenetic regulatory dynamics in models of methamphetamine-use disorder. Genes. 2021;12(10):1614. doi:10.3390/genes12101614. Original commentary on the supplied historical narrative review. Underlying studies are discussed through that review rather than presented as independently examined primary sources.