Wednesday, June 3, 2009

Sensor-based Analyzers

Eclectic, Electric Olfactory

Take a Bite of Flavor Analysis with Electronic Noses and Tongues

In the pharmaceutical field, formulations are usually characterized by three major features: Physical characteristics, such as size, hardness, friability; chemical aspects, such as content and stability of drugs; and sensory attributes, including taste and odor. Among these three, the sensory characteristics are the ones patients initially recognize. In order to improve odor or taste of formulations, flavors are often used in personal care and pharmaceutical products. They are commonly included as an important part of the formulations, especially in pediatric ones, with the aim to mask drug bitterness and/or make the formulations more pleasant.

It is crucial to perform qualitative and quantitative analysis of flavors during formulation development but also during routine production and to study their stability.

Usually, this flavor analysis for product development is conducted by human organoleptic assessment (trained sensory panel). The use of human panelists for odor and taste is accurate, but costly, time-consuming and can be adversely affected by external parameters such as illness or fatigue. This study will focus on the possible use of an Electronic Nose as an alternative tool for organoleptic assessment of flavored oral formulations.

Odor and Taste Evaluation

Electronic instruments for smell and taste analysis have been developed and optimized to meet pharmaceutical needs: increasing formulation candidates in the pre-screening steps, minimizing human test panels and generating objective measurements of taste and reducing formulation development time and costs.

These instruments are sensor-based analyzers that make a global analysis of the total complex chemistry of the sample (chemical fingerprint). They perform qualitative/quantitative analyses of organoleptic and chemical properties of products. The "electronic tongue" is designed for taste analysis in liquid matrix such as gels, syrups, solutions, emulsions, tablets, lozenges, capsules or films, whereas the "electronic nose" analyses a wide range of odors and volatile compounds.

Molecules of the tested sample interact with the active part of the sensor (metal oxide for the gas sensors and chemical sensitive layer for the liquid sensors). These bindings and interactions modify the physical properties of the sensor and generate either resistance or potential variations. These electrical output signals are recorded over time. The signals obtained by each sensor (between six and 18 metal oxide sensor for the "Electronic Nose" and seven liquid sensors for the 'Electronic Tongue') are processed with mathematical models included in software compliant with 21 CFR Part 11.

These analytical instruments provide simple answers, like recognized as "orange flavor, "good conformity" or "no conformity" or more sophisticated responses such as an odor/taste intensity, a molecule concentration, a score of bitterness or a measurement of bitterness-masking efficiency.

Flavor Analysis in an Oral Formulation

The following study describes an application led for the pharmaceutical industry on flavors used in medicines.

Several different samples of six flavors (raspberry, cherry, red berry, pineapple, strawberry and orange) were tested for both qualitative and quantitative studies aimed at:

  • Discrimination of fresh and aged flavor samples;

  • Flavor batch to batch variation analysis;

  • Identification of unknown flavor samples;

  • Quantitation of various flavor concentrations.

Flavor Differentiation

All known samples (A1 to A11) were analyzed with the electronic nose, and data treatment processed through a PCA (Principle Component Analysis). It appears a clear differenciation of the products according to their flavor on a 2-axis chart representing more than 99 percent of the total information (96 percent on PC1 axis and 3.2 percnt on PC2). Reproducibility of results is very good since repetitions of the measurement are quite close one to another for each flavor, except for raspberry (samples A1 to A5). Indeed, the latter includes different batches of the same flavor (A1, A2, A3, A5) and an aged sample (A4), which therefore explains the differences noticed between samples due to batch to batch variations or ageing.

Batch Variation Analysis

Five placebo samples containing different lots of fresh or aged raspberry flavors were analyzed.

  • Samples A2, A3, and A5 show very small batch to batch variations.

  • Samples A1 and A4 (A4 = sample A3 stored at room temperature for 8 months) have significant different fingerprints compared to the other three lots of raspberry flavors (samples A2, A3, and A5).

Explanation for A1: A Gas Chromatography (GC) identification method showed that sample A1 had a different peak profile in comparison with the other three lots (samples A2, A3, and A5). Electronic Nose results correlate well with GC results.

Explanation for A4: Sample A4, obtained by storing the same flavor as A3 during eight months at ambient conditions, was discriminated by Electronic Nose. The instrument can pick up differences between fresh and aged samples. The human nose evaluation of these two samples verified that the raspberry flavor in the aged sample was weaker. The storage time and temperature of formulations might have an impact on the quality of the flavors, and this impact can be detected by electronic nose.

Varying Flavors

In order to visualize the possible differences between samples based on the concentration, a placebo without any flavor and B1 to B5 standards were analyzed. A PCA (Principle Component Analysis) was then processed, showing a clear differentiation of all samples, ranking in a horse shoe shape, which is characteristic of a concentration effect. The Electronic nose thus enables to differentiate all samples and a prediction model for concentrations can be envisaged.

Predicting Model

A calibration curve was generated from these standards (concentrations between 1.01 and 5.02 mg/ml) using a Partial Least Square (PLS) model:

  • X-axis: actual flavor concentrations of the standards input to the model.

  • Y-axis: prediction value produced by the model.

The linear coefficient of determination R = 0.9954 indicates that this calibration model can be used to predict raspberry flavor concentration in unknown samples with high reliability. Moreover, the model was validated by projection of unknown samples and the concentrations predicted by the E-Nose were consistent with the actual values.


In oral formulations, the electronic nose is able to:

  • Distinguish and recognize various unknown flavors (as pure flavoring agents or as a constituent of different formulations);

  • Discriminate different lots of flavor from the same manufacturer;

  • Discern the differences between fresh and aged flavor samples;

  • Quantify flavor amounts.

The electric nose is a rapid tool with an adequate selectivity, sensitivity and reproducibility (measurement precision RSD � 1.3 percent and method precision RSD = 0.5 percent). It can be used to perform flavor identification in pharmaceutical products but also to assay the flavor concentration during release testing, to ensure raw materials quality and freshness according to batches and suppliers and to monitor flavor stability during shelf-life.

The electronic tongue is another innovative tool, especially suitable for a wide range of pharmaceutical applications. This liquid taste analyzer is designed for taste assessment in solution or solids dissolved in liquids. Like the human tongue it provides a global "taste fingerprint" of a complex mixture of organic/inorganic compounds. For all tastes, it performs analysis with similar or better detection threshold than the human tongue. In the pharmaceutical industry, it is used to analyze various oral shapes (tablets, syrups, powders, capsules and lozenges) so as to quantify taste masking efficiency of formulations, analyze medicines stability in terms of taste, benchmark target products or determine the just necessary substance quantity.

It can equipped with a bitterness prediction module (BPM), which consists of a particular set of 7 sensors and a data treatment model correlated with in vivo results. It allows bitterness score prediction of NCE based on a 1 to 20 scale.�


Flavor Quantitation Using Electronic Nose Technology: Application for Stability Analysis in a Pharmaceutical Oral

L. Zhu, R. Seburg, E. Tsai (Merck Research Laboratory, West Point, PA) - S. Isz, J.C. Mifsud, V. Schmitt - ALPHA MOS (Hanover, MD)- Journal of Pharmaceutical and Biomedical Analysis 34 (2004) 453 - 461

Marion Bonnefille, of Alpha MOS Multi Organoleptic Systems (Toulouse, France), can be reached at 33 0 5 62 47 64 55

No comments: