Quantitative characterization of lactose crystalline forms

Quantitative Characterization of Lactose Crystalline Forms by FT-NIR: Application Note Summary

Significance of the topic

Lactose occurs in multiple solid forms (α-lactose monohydrate, α-lactose anhydrous, and amorphous lactose) that influence processing behavior, product performance and stability in pharmaceuticals and foods. Rapid, routine quantification of these forms in raw materials and intermediates supports quality control, process troubleshooting and shelf-life prediction. FT-NIR offers high throughput, minimal sample preparation and potential for use outside the core analytical laboratory, making it attractive for routine crystallinity assessment.

Objectives and overview of the study

- Demonstrate feasibility of FT-NIR to quantitatively determine proportions of the three primary lactose forms in binary and ternary mixtures.
- Develop and validate chemometric calibrations (PLS) for each component using synthetic mixtures and evaluate analytical performance (correlation, RMSEC, RMSEP, limits of quantification).
- Compare NIR-based predictions with orthogonal XRPD data to assess applicability to homogeneous, milled samples.

Methodology and sample preparation

- Materials: α-lactose monohydrate and α-lactose anhydrous (commercial sources); amorphous lactose prepared by spray-drying a 10% aqueous α-lactose monohydrate solution and vacuum drying; final amorphous powder stored over desiccant.
- Sieving: crystalline materials were sieved to 38–125 µm; amorphous material screened to remove >150 µm to avoid agglomeration.
- Mixture design: controlled binary and ternary mixtures prepared by mechanical mixing; ternary mixtures avoided >40 wt% amorphous to limit agglomeration. Calibration/validation split: binary—two-thirds calibration, remainder validation; ternary—three-quarters calibration, remainder validation, with independent sample preparations and no serial dilutions.

Poutuita instrumentatin (Used instrumentation)

- Antaris FT-NIR Solid Sampling System (Thermo Scientific) with integrating sphere for diffuse reflectance measurements through the bottom of standard 2-dram vials.
- Detector: InGaAs.
- FT-NIR acquisition settings: spectral range 4000–12000 cm-1; resolution 4 cm-1; 90 co-averaged scans; acquisition time ~67 s; room temperature (~21 °C).
- Sample preparation hardware: Yamoto Pulvis Mini spray drier for amorphous preparation, vacuum oven (50 °C), mechanical shaker for mixing, sieves for particle size control.
- Chemometrics: Thermo Scientific TQ Analyst software; PLS-I modeling with automated region selection and pre-treatments (Multiplicative Scatter Correction, Norris and Savitzky–Golay derivatives).

Chemometric approach and spectral treatment

- Spectral differences among the three forms are pronounced in FT-NIR and are enhanced using second-derivative preprocessing.
- Region selection was performed per model to target informative overtone and combination bands (examples: 4482–7362 cm-1, 4287–4945 cm-1, 4902–5768 cm-1, 5249–8916 cm-1).
- Pre-treatments included MSC and derivative filters (Norris or Savitzky–Golay) to minimize scatter and baseline variation prior to PLS regression. PRESS plots guided selection of PLS factor numbers.

Main results and discussion

- Binary mixtures: robust calibrations were achieved for all binary combinations. Representative performance:
- Amorphous vs α-lactose monohydrate (α-monohydrate model): r = 0.9992; RMSEC = 1.29; RMSEP = 1.98.
- Amorphous vs α-lactose anhydrous (α-anhydrous model): r = 0.9995; RMSEC = 1.05; RMSEP = 2.12.
- α-lactose anhydrous vs α-lactose monohydrate (α-anhydrous model): r = 0.99999; RMSEC = 0.136; RMSEP = 0.718 (strongest response observed for crystalline–crystalline pair).
- Ternary mixtures: simultaneous quantification of all three forms was feasible but showed reduced precision for the amorphous component relative to the crystalline forms. Example metrics:
- α-lactose monohydrate: r = 0.9981; RMSEC = 1.63; RMSEP = 3.95.
- α-lactose anhydrous: r = 0.9971; RMSEC = 2.04; RMSEP = 3.40.
- Amorphous lactose: r = 0.9817; RMSEC = 2.26; RMSEP = 2.02.
- Limits of quantification in the ternary context ranged approximately from 4.1% to 6.4% (estimated as 3×SD of low-level replicate predictions).
- Comparison with XRPD: application of NIR calibrations developed from binary mixtures to homogeneous, milled sucrose samples (prior study) showed that NIR tracked XRPD trends with a systematic bias that can be corrected; this supports applicability of synthetic-mixture calibrations to real samples with appropriate validation.

Benefits and practical applications

- FT-NIR enables rapid (≈1 min per sample), non-destructive and operator-independent assessment of lactose solid form composition suitable for receiving QC, in-process checks and production-floor monitoring.
- The technique accommodates typical production environments (Antaris instruments designed for robustness) and requires minimal sample preparation compared to XRPD, DSC or NMR.
- Quantitative NIR models allow simultaneous multi-component determination (both crystalline forms and amorphous content), facilitating material fitness-for-use decisions and early detection of contaminated or out-of-spec batches.

Limitations and considerations

- Amorphous lactose exhibits weaker, less distinct NIR responses; this reduces calibration robustness and increases uncertainty relative to crystalline components. Careful calibration design and validation are required where amorphous levels are critical.
- Particle size and agglomeration can influence scatter and spectral baselines; consistent sieving and sample handling improve model performance.
- Synthetic calibration sets can be effective, but application to real-world lots requires orthogonal validation (e.g., XRPD) and, if necessary, bias correction.

Future trends and potential applications

- Transfer of FT-NIR calibrations into routine manufacturing analytics and at-line monitoring for continuous quality verification of excipients and powders.
- Development of robust, transfer-ready calibration families that account for particle size, moisture and batch-to-batch variability to broaden applicability across suppliers and processing states.
- Integration with PAT (process analytical technology) frameworks for real-time control of crystallinity during drying, milling or formulation steps.
- Use of improved detectors, faster FT-NIR hardware and advanced machine-learning regression techniques to further enhance sensitivity to amorphous content and lower LOQs.

Conclusion

FT-NIR combined with multivariate calibration (PLS) provides a practical, rapid and effective approach to quantify α-lactose monohydrate, α-lactose anhydrous and amorphous lactose in binary and ternary mixtures. Crystalline components yield stronger, more reliable calibrations than the amorphous form; nevertheless, achievable LOQs (≈4–6%) and validation against orthogonal methods (XRPD) indicate utility for routine QC and production monitoring when models are properly developed and validated.

References

• Application note authors: Jeffrey Hirsch, W. J. McCarthy, P. E. Luner, A. D. Patel, J. J. Seyer; Thermo Fisher Scientific (Antaris FT-NIR). AN50775_E 0522, Thermo Fisher Scientific Inc., 2022.