Total THC by GC-MS: The Calibration Matrix Can Change the Reported Result

A calibration curve is supposed to connect instrument response to concentration. That connection fails when the calibrants and the samples do not behave the same way inside the analytical system.

This is not a theoretical concern for total-THC testing. In a 2025 Forensic Chemistry study, scientists from the National Institute of Standards and Technology examined the behavior of tetrahydrocannabinolic acid (THCA) and delta-9-tetrahydrocannabinol in a GC-MS system. They identified a plant matrix effect that increased cannabinoid response in cannabis extracts relative to neat calibrant solutions. Under the external-calibration conditions they examined, the mismatch could produce a positive bias in reported total THC.

The instrument was not simply "wrong." The samples and standards were interacting differently with active surfaces in the inlet.

Why total THC is analytically complicated

Total THC generally combines measured delta-9-THC with the THC equivalent of THCA. The acid form loses carbon dioxide when heated, producing neutral THC. The commonly used calculation applies a molecular-weight correction to THCA before adding it to measured delta-9-THC.

Gas chromatography adds another complication because the heated inlet can decarboxylate THCA during analysis. If conversion is incomplete, variable, or misunderstood, the total can be biased. The 2025 NIST study revisited that accepted explanation using isotopically labeled analogs capable of distinguishing material introduced as THCA from material introduced as THC.

The authors reported that THCA was converted to THC at 100 percent efficiency in the plant extracts under the tested conditions, including at relatively low injection temperatures. Their more unexpected finding concerned adsorption. Neat calibrants lost more analyte to active surfaces than the cannabis extracts did. The plant matrix appeared to protect the analytes from those losses, causing sample response to exceed the response expected from neat standards.

For background, see Laboratory Sample Preparation and Matrix Effects in Toxicology.

How adsorption can produce a positive bias

GC inlet liners and other heated surfaces may contain active silanol sites. Polar compounds can interact with those sites instead of traveling efficiently through the system. If a neat standard experiences greater adsorption than an authentic plant extract, equal concentrations no longer produce equal responses.

Imagine that the calibrant loses part of its signal but the plant extract does not. The calibration curve teaches the software that a relatively small response represents a particular concentration. When the better-protected sample produces a larger response, the calculation assigns too much analyte to it.

The problem is comparative. A laboratory could see acceptable peak shape, stable retention time, and a linear calibration while still having a biased comparison between neat calibrants and matrix-containing samples.

That is why method validation must evaluate the complete relationship between calibrants, controls, and authentic matrices. See Calibration in Toxicology and Forensic Toxicology Method Validation.

Why the bias can matter near a classification threshold

The NIST researchers studied three cannabis extracts with total-THC levels near the 0.3 percent threshold used in federal hemp classification. In their experiments, external-standard GC-MS without adequate control of the matrix effect overestimated total THC. The authors reported that the biased approach would have classified the tested hemp samples as marijuana, while other approaches placed them at the threshold.

That result should be read within its experimental boundary. It does not establish that every laboratory using GC-MS misclassifies hemp. It establishes that matrix mismatch can be outcome-determinative when the reported value is close to a legal or regulatory cutoff.

Near a threshold, the important question is not merely whether a number exceeds 0.3 percent. The method must also establish that the number is accurate enough, and its uncertainty controlled enough, to support the classification.

What analyte protectants do

The study evaluated analyte protectants, compounds intended to occupy or block active sites so THCA and THC can pass through the inlet more consistently. Adding the selected protectant mixture to both samples and calibrants reduced the response mismatch and improved the performance of external calibration.

The researchers also found that isotopically labeled internal-standard calibration reduced bias because the analyte and its labeled counterpart experienced similar effects. Other viable approaches discussed in the paper included matrix-matched calibration and standard addition, although standard addition was not evaluated in the study.

No corrective technique is automatic. An analyte protectant must be compatible with the method and must not react with the analyte or create a new interference. A labeled internal standard should be introduced early enough to track the preparation steps it is supposed to control. The authors noted that their labeled internal standards were added after extraction because of concentration and cost constraints.

The bottom line

The reported total depends on more than the mass spectrometer. It depends on whether the laboratory made the calibrants and authentic samples analytically comparable.

The 2025 NIST study showed that active-surface adsorption and protection supplied by the cannabis matrix can change GC-MS response. Under the tested external-calibration conditions, that difference produced positive bias. Analyte protectants, isotopically labeled internal standards, and other matrix-aware approaches reduced or avoided the problem.

A clean peak and a linear curve do not answer the decisive question. Did the standards and the evidence behave the same way?

Source

• Mulloor J, Wilson WB, Sander LC. New perspectives on THCA decarboxylation and accurate GC-MS quantitation of Total THC in Cannabis using analyte protectants. Forensic Chemistry. 2025;44:100668. NIST publication record. doi:10.1016/j.forc.2025.100668.

This article paraphrases the published study. The source PDF, tables, figures, and publisher-formatted pages are not reproduced.