In summary, our findings demonstrate significant increases in coherence length, crystallite size and non-uniform strain of calcifications with increasing tissue malignancy. Values of these characteristics were comparable to other natural apatites; with size ranging from 23 to 39 nm and non-uniform strain in the order of 10−3 [21, 25]. These findings support the view that the mechanisms of calcification formation in different tissue microenvironments may vary, thus below we propose a new model for this process. This is summarised in Fig. 5.
Tissue Microenvironment and Calcification Formation
Tissue microenvironments surrounding benign, in-situ and invasive cells can differ greatly and therefore the ionic milieu within which calcifications form is varied. The ability of an invasive cell to metastasise is governed by the conditions in its microenvironment, which is moderated by numerous factors including pH, proteins, the extracellular matrix, cellular processes and ion concentrations [12]. Biogenic calcifications are also known to possess different characteristics, such as phase composition and carbonate content when associated with different tissue pathologies [14, 15, 22].
Acidity is an important factor for calcification formation. Extracellular acidity is also a key marker for cancer aggressiveness, caused by the switch to a glycolytic respiration type, induced by hypoxia in tumour cells, known as the Warburg effect. Glycolysis produces a higher level of hydrogen ions compared to aerobic respiration, which in turn efflux from the cytoplasm to maintain a neutral or alkaline intracellular pH, causing the extracellular pH to become more acidic [12, 32, 33]. In addition, the action of carbonic anhydrases (CA), intracellularly and extracellularly, leads to an acidic extracellular pH due to the conversion of water carbon dioxide to hydrogen and bicarbonate ions in the extracellular fluid [34]. The expression of one extracellularly acting CA, CAIX, has also been associated with higher DCIS grade and invasive carcinoma, concurrent with an increasingly acidic extracellular pH with malignancy [35, 36].
An impact on calcification formation is reported in studies where an acidic pH increases the likelihood of forming an acidic precursor during the hydroxyapatite formation process [11]. Neutral and alkaline pH favours the direct conversion from ACP to HAp, whereas an acidic pH causes the formation of an additional precursor such as OCP [37]. Therefore, differences in extracellular pH between benign, in-situ and invasive cells would result in different formation mechanisms for hydroxyapatite microcalcifications. These different formation mechanisms lead to the incorporation of different types of carbonate substitution; ACP hydrolysis favouring substitution of carbonate into both the hydroxide site (A-type) and phosphate site (B-type) whereas OCP hydrolysis favours the incorporation of carbonate within the B-type sites [38]. Therefore, benign cell, neutral pH environments promote ACP hydrolysis that is more likely to produce A/B carbonated apatites, whereas invasive cell environments, (favouring OCP hydrolysis), would result in predominantly B-type carbonate apatites. This is further supported by the observations that sodium incorporation into the HAp lattice induces OCP formation, and sodium substitution is highest in invasive calcifications [15, 39]. Hypoxia has also been shown to initiate osteogenic differentiation in other cell types, therefore may contribute to the differing formation mechanisms between benign, in-situ and invasive breast tissue pathologies [40].
Additionally, at low carbonate weight percentages (<4 wt%), a lower carbonate wt% has been associated with B-type carbonate substitution, and a higher wt% favours A-type [20]. Taken together with studies reporting a lower total carbonate level in invasive calcifications and increasing levels with decreasing malignancy, this suggests the favouring of B-type carbonate in invasive calcifications, and A-type in benign [22].
Together the acidic extracellular pH and previously reported carbonate concentrations in microcalcifications suggests a higher level of B-type carbonate in HAp calcifications associated with invasive tissue pathology. Equally, the neutral pH and higher carbonate wt%of benign cells suggest an increased A-type carbonate incorporation. These mechanisms are reflected in the microstructure measurements and lattice parameters measured in benign, in-situ and invasive calcifications.
Calcification Microstructure
Previous studies have demonstrated a decrease in total carbonate substitution with increasing tissue malignancy, which should lead to an increase in crystallite size with malignancy [17, 22]. This observation is also supported by our data; microcalcifications associated with invading tumours were characterised to possess the largest crystallite size (Fig. 4) For the first time, our analysis also enables a more specific inference regarding the lattice carbonate distribution.
Increasing total carbonate substitution has also been linked to an increase in non-uniform strain, therefore a decrease in non-uniform strain with malignancy would be expected [21]. However, the data presented here indicates the opposite trend. B-type substitutions have previously been shown to increase non-uniform strain [17]. B-type substitutions will, in principle, for the same number of substitutions, produce greater non-uniform strain than that of A-type. This is because carbonate substitution for phosphate occurs randomly at 6 crystallographically equivalent sites (hm) within the unit cell, compared to 2 (e3) for hydroxide substitutions. The non-uniform strain values observed within this study, suggests an increasing B-type carbonate substitution with increasing malignancy. However, given the total carbonate decreases with malignancy, this study suggests that lattice carbonate of breast microcalcifications occupies both A-type and B-type sites [22].
This model is further supported by previous estimates of apatite unit cell dimensions as revealed by lattice parameter measurements. A-type carbonate substitutions increase the apatite ‘a’ axis due to CO3 having a significantly greater ionic radius than OH. In contrast, B-type carbonate decreases the ‘a’ axis due to CO3 having a smaller radius than phosphate [28]. In addition, A-type carbonate substitutions decreases the ‘c’ axis and B-type increases the ‘c’ axis [28]. Previous studies have shown a decrease in the ‘a’ axis with malignancy and an increase in the ‘c’ axis, further supporting the view that the quantity of B-type carbonate substitution increases and A-type substitution decreases with increasing malignancy [14].
The distribution of data observed in all the Williamson-Hall plots is indicative of microstructural anisotropy. The lack of reliable peaks in reflections other than the 00ℓ means that only data in this reflection could be separated into size and non-uniform strain. However, the noted increases in CL with malignancy in all three reflections analysed (Fig. 3) suggest that crystallite size and/or non-uniform strain are also changing in the <hk0> and < 0 k0> directions. Proteins have been noted to interact with the non-basal faces of HAp, restricting growth in neutral pH [41]. Extracellular acidity may cause changes in protein conformation in cancer cells permitting increased CL in the <0 k0> and < hk0> directions, hence a directional variance in difference magnitude. Thus, for the development of quantitative biomarkers based upon microstructural features, appropriate choice of lattice direction to maximise sensitivity is required.
The Importance of Microstructural Differences
Studies have previously shown that the total carbonate content of hydroxyapatite is positively correlated with the solubility of the crystals, which would suggest benign breast calcifications are more soluble than pathological ones as a consequence of their respective carbonate contents [21, 22]. However, more stable calcifications are generally considered to be benign [42]. A positive correlation between non-uniform strain and carbonated apatite solubility has previously been found, which may offer an explanation for this apparent contradiction, where less strained crystallites, i.e. those in benign calcifications, are less soluble [21]. This may suggest that the type of carbonate also has a role to play in other calcification characteristics. In addition, a more acidic pH increases the solubility of hydroxyapatite, which may also play a role in the relative instability of calcifications in pathological tissue [43].
Furthermore, the interaction of calcification with proteins in the surrounding tissue has been shown to play an important role in the propagation of tumour virulence characteristics. For example, the presence of hydroxyapatite in mammary cell lines has been shown to upregulate matrix metalloproteinases which are key proteins in the degradation of the basement membrane, enhance mitogenesis, and induce the production of interleukins (ILs) [44]. It has also been demonstrated that the surface morphology of hydroxyapatite crystals impacts the level of protein adsorption on to the crystals, which can have downstream effects on expression of osteoblastic markers such as Runt-related transcription factor 2 (RUNX2) and alkaline phosphatase (ALP) [45]. Moreover, the substitution of ions such as carbonate into the hydroxyapatite lattice can affect crystal morphology, meaning that the microstructure of microcalcifications potentially has the ability to impact downstream effectors and hence tumour virulence characteristics [20, 46].