Optimizing rapid identification of prohibited substances in sport using ASAP-MS and an advanced data-analysis workflow

Optimizing rapid identification of prohibited substances in sport using ASAP-MS and an advanced data-analysis workflow

Alisha Henderson¹, Oliver Krug², Ashley Sage³, David Douce³, Scott J. Campbell⁴, John Moncur⁴, Mario Thevis², Liam M. Heaney¹

¹Loughborough University, Loughborough, UK. ²German Sport University Cologne, Germany. ³Waters Corporation, UK. ⁴SpectralWorks Limited, Runcorn, UK.

Correspondence: Liam M. Heaney (L.M.Heaney2@Lboro.ac.uk)

First presented 74^th ASMS Conference, June 2026, San Diego, USA.

Targeted Compound Identification

• Ambient MS techniques offer options for rapid and direct analyses to benefit sports drug testing as a complimentary technique in anti-doping control^1-3
• Black market pharmaceutical products known to contain prohibited substances in sport were analysed using a Waters RADIAN ASAP-MS and SpectralWorks AnalyzerPro XD • • A compound library was developed within AnalyzerPro XD using standards of drugs known to be present within the products
• Batch tested (clean) sports supplements were analysed to assess the capacity for the search function to eliminate background ions from causing false positive results

Combating False Positives

• Initial iterations of software-driven ID matching observed a repeated false positive ID for trenbolone
• The [M+H]+ ion for trenbolone (m/z 271) matched a fragment ion formed from the cleavage of multiple T ester compounds (Fig 1)
• Library matching criteria were optimised to reduce the impact of the variable presence of m/z 271 across data channels
• Alterations of the forward:reverse confidence matching ratio and addition of ion ratio values for the two most abundant ions in each channel did not solve the issue
• Increasing the reverse match score to 700 improved the comparison of the library to the unknown sample and removed the false positive ID
• This change was applied as bespoke for trenbolone as it reduced success for other compound IDs

Figure 1. False positive ID for trenbolone

Reducing False Negatives

• Software-driven ID matching also reported false negative ID outcomes for T propionate
• The presence of multiple T Ester drugs caused significant perturbation of m/z 97 and 109 fragment ion intensities (Fig 2)

Figure 2. False negative ID for T propionate

• A further reporting feature of a heatmap of the isolated precursor ion in data channel 1 (m/z 345) provided clear identification of the samples containing T Propionate against all other samples (Fig 3)

Figure 3. Heatmap showing presence of m/z 345 (T propionate)

Identifying Suspect Samples

• The ability to identify suspect samples that might contain sports prohibited substances would offer excellent value in addition to targeted compound screening
• A series of pilot investigations to assess the ability to ‘flag’ suspicious compounds were evaluated

Projection of Global Ion Profiles

• A PCA model was created using known prohibited substance standards classified as ‘anabolic steroids’ or ‘other drug’
• Black market pharmaceutical products and batch tested sports supplements were projected on to the model to assess rapid visual classification as either an anabolic steroid, any other prohibited drug, or a clean supplement
• The model performed poorly (Fig 4) with only 56% of unknown samples being categorised correctly, yielding a sensitivity of 64%, a specificity of 50%, a positive predictive value of 47%, and a negative predictive value of 67%
• The data demonstrated that the model performed more appropriately to identify samples not related to the model build (i.e. the clean supplements) than it did to identify samples containing the modelled drugs

Figure 4. PCA model of prohibited substance standards and projected predictions for pharmaceutical products and clean supplements

Flagging of Suspected Testosterone-based Drugs

• An independent library based on the fragment ions from testosterone was developed
• The library entry required the presence and pattern of fragment ions (predominantly m/z 97 and 109) in the highest cone voltage (50V) data channel with a high base peak ion threshold
• The library entry was able to correctly flag all samples suspected to contain a testosterone-based drug (Fig 5a) without any false positive outcomes for alternative drugs (Fig 5a) or batch tested supplements (Fig 5b)

Figure 5a and 5b. Flagging of suspect T-based drugs

PDF download is available here.