兽研摘要
牙结石是一种广泛存在于家犬口腔中的矿化口腔生物膜，但其矿物学变异性和化学演化特征仍未被充分阐明。本研究以28只不同品种、年龄和临床背景的室内宠物犬为对象，采集其常规专业洁牙过程中作为生物废弃物的牙结石样本。研究联合使用X射线粉末衍射、振动光谱和元素分析，测定沉积物的晶相组合、磷酸盐和碳酸盐光谱特征以及主量元素组成。晶质部分以羟基磷灰石和镁方解石为主，其组成范围广泛，从富方解石型到富磷灰石型沉积物均有分布。对部分富磷灰石样品的拉曼分析进一步表明，存在结构无序的磷酸盐贡献，与非晶态磷酸钙相符；对代表性富磷灰石样品的定量物相分析则显示存在相当比例的非晶质组分。元素数据显示，磷含量随晶质羟基磷灰石比例单调增加，而钙和镁在整个样品集中保持相对恒定。Ca/P比值明显高于化学计量羟基磷灰石，并随羟基磷灰石比例增加呈非线性下降，这支持存在含钙的非磷灰石组分，并证实主量元素化学可作为方解石—磷灰石平衡的独立描述指标。相关分析和探索性多变量分析表明，羟基磷灰石富集与犬年龄、去除的牙结石质量和估计累积时间呈中度关联，界定了一个与年龄和累积相关的矿化连续谱，而非固定的牙结石组成。这些结果为犬牙结石作为一种异质性生物矿物系统提供了矿物学和化学框架，并提示其组成记录了生物膜矿化、沉积物累积和矿物成熟等耦合过程。本研究为横断面研究，未验证诊断用途，但为未来对犬口腔健康中牙结石矿化的受控和纵向研究奠定了基础。
[内容请查看原文]
引言
牙菌斑是动物中最常见且研究最广泛的生物膜之一。其发展始于唾液蛋白吸附于牙齿表面以及先锋微生物的黏附，随后发生细菌间共黏附和微生物演替，形成更复杂的群落。当局部液体相对于磷酸钙盐达到过饱和时，牙菌斑发生矿化并形成牙结石（DC，也称牙垢），这一条件受pH和产碱细菌代谢的强烈影响。矿物成核可异相发生于细菌细胞表面、膜磷脂和蛋白脂质、细胞外基质成分、膜衍生囊泡和细胞碎片上。唾液蛋白、焦磷酸盐、镁及其他促进剂或抑制剂可进一步调控成核、晶体生长和相选择。机械去除牙生物膜是人类预防生物膜细菌产生的酸和毒素攻击牙釉质、损害牙周组织并诱发牙龈炎症（最终导致牙齿脱落）的常见做法。这可在人类中防止大量DC形成，而在宠物（如家犬）中，定期洁牙不切实际，大量DC形成相当常见。
全球家犬种群持续增长，目前估计已达五亿只，近期研究已充分评估宠物犬口腔疾病的高患病率。最常见的口腔疾病是牙周病（PD），而DC是前者的固有组成部分。普遍认为，炎症及由此产生的组织损伤是由于对细菌感染的免疫反应调控不当所致。鉴于DC与全身健康之间的潜在联系，科学工作应首先解决其物理化学特征，然后探讨其与犬健康状态的关系。
人类DC中已鉴定出多种磷酸钙相，包括透钙磷石或二水磷酸氢钙、磷酸八钙、含镁白磷钙石和含碳酸盐羟基磷灰石。透钙磷石主要与早期龈上沉积物相关，而磷酸八钙、羟基磷灰石和白磷钙石已在龈上和龈下牙结石中均有报道。当前模型提出，牙结石矿化可能最初涉及非晶态或低有序磷酸钙以及动力学有利的前驱相（如透钙磷石和磷酸八钙），随后水解并成熟为溶解度更低的羟基磷灰石和白磷钙石。这一演化不应被视为唯一的线性途径，因为局部pH、液体组成、镁、氟化物和大分子抑制剂可促进或稳定不同的相。
分析牙结石最实用的方法是测定其矿物组成。LeGeros与Shannon率先开展了犬牙结石结晶组成的研究，分析了约十只个体的样本。他们通过X射线方法表明，其主要成分为方解石形式的碳酸钙和磷灰石形式的磷酸钙。由于犬唾液中碳酸盐/磷酸盐浓度的差异，观察到方解石/磷灰石比值因犬而异。该比值在牙结石样本的最内层高于最外层，提示在犬唾液磷酸盐存在的情况下，方解石易于水解为磷灰石。方解石占主导并混有少量磷灰石，是犬唾液中高pH和高碳酸盐浓度条件以及蛋白水解活性诱导的静息菌斑pH较高共同作用的结果。事实上，犬唾液相对于碳酸钙处于过饱和状态，而磷酸根阴离子选择以方解石多晶型形式沉淀碳酸钙。
随后，LeGeros等人评估了扫描电子显微镜（SEM）和电子探针显微分析在揭示人类和动物（包括犬）牙结石固体成分方面的效力。该研究报告了Ca、Mg的定量分析和Ca/P比值，显示动物结石的Ca/P摩尔比高于2.5（人类为1.3–1.7），且Mg含量与Ca/P比值不相关。高于15的比值被归因于完全由方解石组成的区域。
近期，Chiba-Ohkuma等人对七只老年室内犬（>15岁）进行了多方法学分析，证实犬结石的Ca/P比值高于人类结石，且有机物比例更大，这归因于犬唾液中没有淀粉酶。X射线衍射分析提示为方解石和含碳酸盐羟基磷灰石的混合物。
Nakashima等人从犬牙结石比人类牙结石更易脱落的证据出发，分析了釉质/结石界面的形态和组成。他们显示存在光滑表面和100 μm宽的间隙，表明粘附力差。微量分析显示，与釉质接触的表面Ca和P含量降低，但不足以确定其是富方解石还是富磷酸盐。结合Borah等人 reported 的研究——该研究确立犬唾液的无机成分限制碳酸化羟基磷灰石的沉淀——作者声称需要进一步研究以弥合犬牙结石组成中的矿物学空白，特别是由于钙盐对牙结石形成的关键作用。
离子化形式的Ca在生化反应中最为重要。更严重的PD分期通常与较低的血清Ca2+值相关。有趣的是，年龄、血清Ca2+水平与PD分期之间观察到统计学显著相关性，而与性别或体重则未建立此类相关性。PD在约80–85%的2岁以上犬中表现，该病影响几乎所有5岁以上的动物，老年犬表现出更晚期的疾病阶段，因为PD分期随着患者年龄增长而恶化。
在本研究中，我们结合定量矿物学分析、主量元素测定和探索性统计分析，对采集自28只犬群体的牙结石进行了研究。结果表明，犬牙结石并非具有固定的矿物学组成，而是跨越一个宽广的方解石–磷灰石连续体。这种组成变异性定义了一条矿物学成熟/积累轨迹，从富方解石沉积（被解释为相对早期阶段或较不成熟的矿化产物）到富磷灰石沉积（代表更晚期的矿化状态）。
[内容请查看原文]
材料与方法
[内容请查看原文]
牙结石采样
在常规专业牙齿清洁过程中，由持证兽医（意大利米兰Clinica Veterinaria Bicocca）在全身麻醉下使用超声波洁牙机对28只不同品种的室内犬进行口腔卫生和疾病预防操作，将产生的生物废弃物作为牙结石样本获取。收集的材料包括龈上和龈下牙结石。由于常规超声波洁治会逐步从两个部位剥离材料，在临床操作过程中对两种组分进行可靠分离实际上不可行。因此，将其作为混合牙结石进行分析。据此，本研究探讨的是专业牙齿清洁过程中去除材料的总体矿物学组成，并未区分龈上和龈下沉积物之间的部位特异性差异。当兽医临床指示需要进行专业牙齿清洁且获得主人同意对去除材料进行匿名化科学使用时，犬只被机会性纳入。未根据品种、性别、繁殖状态或预期矿物学组成进行选择。未为研究目的进行超出标准兽医护理的任何操作。研究方案由Clinica Veterinaria Bicocca（意大利米兰）的动物福利机构（Organismo Preposto al Benessere Animale，OPBA；Animal Welfare Body，AWB）评估，该机构确认根据第26/2014号立法法令和2010/63/EU指令无需正式伦理批准，因为研究仅涉及常规兽医牙科操作中获得的生物废弃材料，未对活体动物进行任何实验性干预。本研究按照ARRIVE指南（https://arriveguidelines.org）报告。本研究未对任何动物实施安乐死。在临床操作和样本采集之前，已从所有犬主获得书面知情同意。主人被告知专业清洁过程中去除的部分材料可经匿名化用于矿物组成的科学分析。所有样本均按照GDPR合规的匿名化方案处理，以确保无法检索到关于犬只或主人的任何身份信息。
临床信息从常规兽医记录中检索。未为研究目的执行包括系统性牙周探诊和口内X线摄影的标准化牙周分期方案。因此，此处使用的口腔健康评分（见第节）应被视为基于所报告临床观察的探索性序数描述符，而非正式的牙周诊断。对每只犬记录以下信息：ID、年龄、品种、体重、性别和繁殖状态、既往牙齿清洁史、饲养习惯、 reported病理和外科手术，以及经常活动的主要户外环境。牙结石样本称重（收集的牙结石样本质量范围为10至800 mg），用去离子水充分冲洗后在空气中干燥24小时。由于常规超声波洁治过程中的收集并非设计用于定量回收全部牙结石负荷，该变量代表为分析收集的材料质量，而非口腔中存在的牙结石总量。因此，它被用作牙结石积累的操作性替代指标，不应被解释为总牙结石负荷的精确测量。样本代码（如MNU、MLU、GMA等）指匿名化的个体犬只标识符。与被调查犬只相关的收集信息列于表S1中。
样本在研钵中研磨成粉末并储存在干燥器中。后续分析使用以下方法进行。
[内容请查看原文]
X射线粉末衍射（XRPD）
使用X‘Pert Pro X射线衍射仪（Malvern Panalytical），采用Bragg–Brentano几何构型（镜面扫描）和Cu靶（λ = 1.54 Å）作为X射线源，在粉末样本上收集衍射图谱，以确定矿物组成和相对含量。鉴于部分犬只可用样本量较少，且为对所有样本采用统一方法学途径以实现一致比较，数据收集使用低背景平面样品架进行，从而能够测量极少量样本。然而，尽管该技术可提供适合可靠测定方解石和磷灰石比例的数据集，但无法用于测定伴随结晶相的无定形含量，后者需要与已知量的内标（刚玉）混合，使该混合样本不再适合后续分析。相反，对于代表性大体积样本（LLI，质量 = 320 mg），还使用刚玉内标进行了数据收集，从而能够对无定形组分进行定量相分析（QPA）（见支持信息）。QPA通过Rietveld精修使用Profex v. 5.6.1进行。
[内容请查看原文]
拉曼光谱
显微拉曼测量使用HR800 Horiba Jobyn-Yvon仪器进行，采用50×物镜（0.75 NA，激光光斑尺寸约1 μm），激发波长为405 nm，由Integrated Optics生产的固态激光器提供。测量前，将样品粉末轻轻压于显微镜玻璃载玻片上。对于粉末样品上的每个不同位点，在采集拉曼光谱前重复激光聚焦程序。样品上的不同位点是在直径数毫米的区域内随机选择的。对于每个粉末样品，我们通过将激光聚焦于可用材料的不同位点来采集多条光谱。我们观察到各位点间的差异可忽略不计；因此，所报告的光谱可视为代表每个样品。所有测量中入射到样品上的激光功率均保持在10 mW以下以避免光损伤。
[内容请查看原文]
主要元素的定量分析
粉末样品在4 mL 65% HNO3、4 mL 37% HCl和2 mL 40% H2O2的混合液中，借助微波消解仪（MV 5000，Anton Paar）于200 °C热处理10 min进行溶解。溶解后结石中Ca、Mg和P的浓度使用电感耦合等离子体-光学发射光谱（ICP-OES）Optima 7000 DV（PerkinElmer，意大利）系统测定。
[内容请查看原文]
统计分析
通过使用匿名化样品ID将每个DC样品与相应的犬只记录匹配，构建统一数据集。用于统计分析的矿物学描述符为通过Profex Rietveld精修获得的羟基磷灰石分数，以下表示为HA%。方解石分数计算为CA% = 100 – HA%，仅考虑XRPD解析出的结晶部分。其他定量变量包括消融时犬龄、体重、消融结石质量、距上次牙齿清洁的时间以及ICP-OES测定的Ca、Mg和P元素浓度。Ca/Mg和Ca/P重量比由相应的元素浓度计算得出。
距上次牙齿清洁的时间定义为：对于接受首次牙齿清洁的犬只，为出生至采样消融之间的月数；对于曾接受一次或多次牙齿清洁程序的犬只，为上次牙齿清洁至采样消融之间的月数。因此，该变量被用作牙结石沉积有效累积时间的估计值。
报告的口腔状况被转换为序数口腔健康评分用于探索性统计分析。无报告全身性或口腔病变的犬只被赋值为0分。患有牙龈炎（包括轻度、中度和中度牙龈炎）的犬只被赋值为1分。患有中度牙周炎的犬只被赋值为2分。患有牙周炎的犬只被赋值为3分。患有重度牙周炎和/或牙齿松动的犬只被赋值为4分。口腔健康信息不可用的犬只被排除在涉及口腔健康评分的分析之外。全身性状况若与口腔状况分开报告，则不纳入口腔健康评分。由于该评分并非正式验证的牙周分期，其与其他参数（如矿相分数）的关系不具有预测意义。
定量变量之间的单变量关联使用Pearson相关系数（r）评估线性关系，以及Spearman秩相关系数（ρ）评估对非线性和离群值较不敏感的单调关系。对选定的关系进行了简单线性回归，包括HA%作为犬龄的函数、HA%作为消融结石质量的函数、消融结石质量作为距上次牙齿清洁时间的函数，以及Ca/P作为HA%或其倒数变换的函数。倒数变换100/HA%也被考虑在内，因为Ca/P比预期会随含磷酸盐结晶相的相对丰度呈非线性变化。
使用Python 3及pandas、scipy、statsmodels和scikit-learn包进行探索性多变量分析，以评估犬DC的成分变异性是否可由年龄、累积和组成相关的耦合描述符来描述。使用statsmodels拟合多元线性回归模型，以HA%为因变量，以犬龄、距上次牙齿清洁的时间、消融结石质量以及适当时的口腔健康评分为解释变量。计算方差膨胀因子以评估预测变量间的共线性。
对标准化变量进行主成分分析，以识别数据集中潜在的协变轴。在将所有犬只均可获得的核心定量变量（即犬龄、距上次洁牙的时间、磨除牙结石质量和HA%）进行标准化后，使用scikit-learn进行了第一次主成分分析（PCA）。第二次PCA在具有完整化学数据的样本子集上进行，包括HA%、100/HA%、犬龄、距上次洁牙的时间、磨除牙结石质量、Ca%、Mg%、P%和Ca/P。PCA作为探索性工具而非预测模型使用。通过检查得分和载荷来识别成分趋势、累积相关趋势以及偏离主要年龄-累积-成分轴的样本。
当变量可用时，所有统计分析均在完整的28只犬数据集上进行。涉及口腔健康评分的分析仅限于具有已知口腔健康信息的犬只。统计学显著性设定为p < 0.05。鉴于队列规模有限且多变量分析具有探索性，结果主要从效应量、关联方向以及变量间一致性的角度进行解读，而非作为确定的因果证据。
[内容请查看原文]
结果
[内容请查看原文]
牙结石的矿物学表征
图 中报告了对代表性且特别巨大的样本（LLI）进行的准确定量XRPD测量。通过考虑羟基磷灰石（HA，Ca5(PO4)3(OH)）和镁方解石[CA，(Ca,Mg)CO3]的存在，完成了衍射峰的完全指标化。相关峰位、hkl指数和主要峰的相对强度报告于表S2a中。HA相相对较宽的峰表明其结晶度低于CA。因此，除HA和CA两种矿物外，非晶相占样本的比例高达61.4%（图S1和表S2b）。
[图：样本LLI。实验（黑色）与计算（红色）衍射图谱及相关强度差（灰色）。主要]
在87%的病例中，镁方解石被建议为最高匹配模式（见支持信息）。根据精修a参数估计的MgCO3量在0.86至2.33 mol%之间，平均为1.16 mol%（表S3）。在此基础上，该矿物可定义为低镁方解石。
在确定这三种物相组成犬牙结石后，可通过Rietveld精修比较相应XRPD反射的强度来估计它们的相对含量。
图 报告了所选XRPD图谱——为提高可读性，仅显示系列中每隔一个的图谱。在等数量不同样本上采集的完整28个XRPD图谱系列报告于支持信息图S2中。可以看出，HA和CA的含量跨越所有比例，从几乎纯磷灰石到几乎纯方解石（图 ），测得的HA%分布呈现增加趋势（图S3）。对图 的目视检查证实了QPA的可靠性。实际上，两种结晶相的主要诊断峰在整个系列中一致变化，仅有少数例外，即PTA，其可能应占据较低位置（另见图S2中的PTO）。然而，应注意PTA和PTO均属于图 中以平台区为特征的HA最富集区域，其中HA与CA之间的相对比例在实验误差范围内变化（表S3）。
[图：DC的选定XRPD图谱——仅选择系列中每隔一个的图谱——显示连续变化]
[图：完整样本结晶部分中HA%和CA%的变化及估计标准差]
拉曼光谱用作互补的相敏探针，以支持含碳酸盐和含磷酸盐成分的鉴定，并通过分析磷酸盐ν1谱带评估磷酸钙部分的结构有序度。LLI样本获得的结果报告于图 中，覆盖200至2000 cm–1的光谱范围。从宽荧光背景中清晰显现的最强峰归属于磷酸盐（约960 cm–1）和碳酸盐（约1070 cm–1）。此外，还可观察到与磷酸盐弯曲模式相关的较弱特征（约400和600 cm–1），以及两个较弱的谱带（CH2摇摆和酰胺I），归因于生物源性羟基磷灰石中的蛋白质。此外，1004 cm–1处的弱峰可初步归因于苯丙氨酸。
[图：LLI样本的拉曼光谱（405 nm激发，样本处功率约10 mW，50×物镜）。黑色箭头指示]
进一步观察发现，磷酸盐谱带呈现结构特征，并显示两个主要组分：一个位于950 cm–1，归属于样品中的无定形部分；另一个位于960 cm–1，归属于结晶部分（图S4）。实际上，HA结晶相通常来源于一种有序度较低、亚稳态的前驱体，称为无定形磷酸钙（ACP，Ca x (PO4) y ·nH2O；n = 3–4.5）。ACP和HA的相对含量可通过拉曼光谱有效监测，因为磷酸盐对称伸缩振动谱带从约950 cm–1（无定形相）移至960 cm–1（结晶羟基磷灰石），同时峰逐渐变窄。该诊断性行为在参考文献中有描述，为表征无定形和结晶磷酸盐组分提供了可靠的光谱学方法。对每个光谱，选取磷酸盐谱带区域（900–1020 cm–1），将信号解卷积为两个洛伦兹组分，分别 centered at ∼950 cm–1 和 ∼960 cm–1，对应于无定形（A）和结晶（C）磷酸盐贡献。扣除线性基线，按照参考文献中概述的步骤进行拟合，R 2 > 0.98。然后以无定形和结晶组分面积之比（A/C = A950/A960）作为各标本中相对无定形含量的半定量指标。个别拟合结果在图S4中展示于富HA样品LLI、LPE、SIA和MRA，所有光谱的拟合结果总结于表S4。四个样品间A/C比值的分布以柱状图形式示于图S5。数据表明磷酸盐有序度存在一些差异；例如，与SIA和LPE样品相比，LLI样品表现出更无序、结晶度更低的磷酸盐结构。在LLI样品中，拉曼分析使我们能够得出结论：在总Ca-磷酸盐中，31%由ACP构成。由于LLI中结晶HA的重量分数估计为37%（表S2b），推算ACP的重量分数为17%。剩余的无定形部分——由Rietveld精修估计的总量约为61.4%——来自其他相，可能为生物成因（见下文）。样品LPE、SIA和MRA中Ca-磷酸盐的ACP分数分别为17%、20%和25%（表S4）。
如图S6所示，约1070 cm–1处的碳酸盐谱带实际上包含两个不同的贡献：一个位于约1072 cm–1，与羟基磷灰石晶格中取代磷酸盐的碳酸根离子（B型）或无定形碳酸钙（ACC）相关，另一个来自与镁方解石相关的碳酸根离子（约1086 cm–1）。为探究通过拉曼光谱定性评估所有碳酸盐和所有磷酸盐基团相对量变化的可能性，我们采集了LLI、SIA、LPE和MRA样品在900–1200 cm–1区域的光谱。我们在线性基线上进行光谱拟合，以评估碳酸盐谱带（约1070 cm–1）的面积，并将其（碳酸盐/磷酸盐比值）与磷酸盐谱带（约960 cm–1）的面积进行比较。光谱拟合结果示于图S7a，而碳酸盐与磷酸盐谱带面积比示于图S7b。我们观察到，碳酸盐/磷酸盐比值沿LLI、SIA、LPE、MRA序列从0.45单调递减至0.20（图S7b），而由XRD数据获得的CA/HA比值几乎恒定（图 ）。然而，必须强调使用这两种技术获得的两个比值之间的根本差异：在XRD中，该比值代表CA和HA结晶相的相对比例，而在拉曼光谱中，它反映的是碳酸根和磷酸根阴离子的相对量。
[内容请查看原文]
牙结石矿物学组成与元素组成之间相关性的评估——单变量统计分析
可以预期一些相关性将矿物学组成（HA和CA的相对分数）与元素组成联系起来。假设组成仅为CA和HA，Ca在CA和HA中均贡献40%w/w，P在HA中贡献18.5%w/w，则预期存在以下关系：
消融牙结石的主要元素和Mg含量以重量百分比表示，作为磷灰石含量的函数示于图 。钙是最丰富的元素，分数为28% ± 9%，而Mg含量设定为0.34% ± 0.12%。值得注意的是，Ca%和Mg%与磷灰石含量不相关，而P%随磷灰石含量从0.5%单调但非线性增加（Pearson's r = 0；Spearman ρ = 0.536，p = 0.0122，表 ）。
[图：所分析c中Ca（红点，左轴）、Mg（绿点，右轴）和P（蓝点，右轴）的重量百分比]
1. Estimated Correlations between HA% and Elemental Composition of Canine Calculi
| Relation | test | coefficient | p -value | comment |
|---|---|---|---|---|
| P% vs HA% | Spearman | ρ = 0.536 | p = 0.0122 | significant, monotone |
| Ca/P vs 1/HA% | Pearson | r = 0.713 | p = 0.000285 | Strong and significant |
图a显示了Ca/Mg和Ca/P比值随磷灰石含量的变化。虽然Ca/Mg比值与HA%无相关性（其值设定为87 ± 26），但Ca/P比值随HA%呈下降趋势，与1/HA%呈强线性相关（图b；r = 0.713，p = 0.000285，表）。考虑到纯HA预期的Ca/P重量比和P%分别为2.16和18.5，而相应的实验估计值为13和1.5，则必须有相当一部分Ca存在于结晶HA之外，包括XRPD鉴定的结晶碳酸盐相以及潜在的非晶态含钙组分。鉴于拉曼光谱中观察到的1072 cm–1强谱带（图和S6），这部分非晶相可能包含ACC。
[图：(a) 所分析结石中Ca/Mg（黑色圆圈）和Ca/P（红色圆圈）的重量比随其HA含量的变化]
[内容请查看原文]
结石与犬特异性之间相关性的评估——单变量统计分析
最强的相关性 observed 于消融质量与距上次消融的年龄/月数之间（图）。
[图：消融结石质量随(a)犬年龄和(b)距上次消融月数的变化。红色实线代表]
消融结石质量与距上次牙齿清洁所经过时间的相关性比与实足年龄的相关性更强，表明收集的结石量与沉积物估计积累时间相关。
关于矿物组成与犬特异性之间可能的相关性，HA%与消融结石质量（表和图S8）、距上次消融的时间（如有，图S9）及犬年龄（表和图）之间也评估出显著但中等程度的相关性。与HA%未显示显著相关性的变量（连同第节中提到的Ca%和Mg%及Ca/Mg）是犬体重。
2. Estimated Correlations between HA% and Some Dog Specificities
| Relation | test | coefficient | p -value | comment |
|---|---|---|---|---|
| Ablated mass vs HA% | Pearson | r = 0.497 | p = 0.0071 | significant |
| Spearman | ρ = 0.548 | p = 0.0025 | significant | |
| Months from last ablation vs HA% | Pearson | r = 0.407 | p = 0.0315 | significant |
| Spearman | ρ = 0.455 | p = 0.0150 | significant | |
| Age vs HA% | Pearson | r = 0.455 | p = 0.0149 | significant |
| Spearman | ρ = 0.403 | p = 0.0333 | significant |
[图：HA的重量百分比随犬年龄的变化。粗红色虚线代表]
相对于HA%-年龄关系的主要离群值包括五只结石中CA%异常高的犬（AGY、TEO、SIA2、WSP和HIE），以及一只HA%异常高的犬（PTA）。富含方解石的负向离群值的特征是消融结石质量相对较低（表），以及轻度至中度口腔疾病（图）。
3. Distribution of Ablated Calculus Mass among the Considered 28 Dogs Divided in a Group of Calcitic Outliers (See Figure ) and All Other Dogs
| Group | n | Average mass (mg) | Median mass (mg) | Range (mg) |
|---|---|---|---|---|
| Calcitic outliers | 5 | 110 | 110 | 30–170 |
| others | 23 | 277 | 240 | 10–800 |
为扩大正向离群值组并平衡比较，在PTA之外还可纳入AES、MRA和SIA（图）。这些离群值分别与相对较高的消融质量160、160和800 mg相关，其口腔健康与重度牙周炎或轻度牙周炎相关（图）。在图中的离群值中，只有AES也表现为图b中的离群值。这意味着HA含量相对于年龄的强烈偏差与异常Ca/P比值无关。
由于HA%也随消融质量增加（图S8，表），HA富集的表观年龄依赖性（图）应解释为反映犬年龄、结石积累历史和矿物成熟的综合效应。
[内容请查看原文]
结石与犬特异性之间相关性的评估——多变量统计分析
探索性多变量分析进一步支持了年龄-积累-组成耦合趋势的存在。以犬年龄、距上次牙齿清洁所经过时间和消融结石质量作为XRPD衍生结晶HA分数预测变量的多元线性回归模型解释了约29%的HA%方差（R² = 0.287；校正R² = 0.198；模型p = 0.0405）。然而，当同时考虑时，没有一个单独预测变量保持统计学显著性（表S5），表明年龄、积累时间和消融质量是部分相互依赖的描述性指标，而非矿物组成的独立决定因素。方差膨胀因子保持中等水平，支持将该模型用作协变异的探索性描述。当在有口腔信息可用犬的子集中将序数口腔健康评分加入模型时，解释方差仅略有增加且模型无统计学显著性，证实可用的临床描述指标应谨慎解释，而非作为正式的牙周分期。
对全部犬只可获得的四个核心定量变量（年龄、距上次洁牙的时间、消融质量和HA%）进行主成分分析，识别出一个主导的第一主成分，解释了总方差的68.4%。该成分对全部四个变量均显示正载荷，与共同的年龄–蓄积–矿物成熟轴一致（表S6）。第二主成分解释了17.0%的方差，主要将HA%相对较高的样本与年龄或蓄积时间相对较高但HA%较低的样本区分开来。因此，这第二个轴突出了偏离主要成熟趋势的样本，包括相对于其年龄或蓄积历史而言HA%较低且富含方解石的样本，以及尽管蓄积相关描述指标有限但HA%相对较高且富含磷灰石的样本。对具有完整化学数据的子集进行的第二次PCA显示，HA%、100/HA%、Ca/P和P%对成分方差贡献显著，与主量元素化学提供方解石–磷灰石平衡的独立描述指标相一致。
[内容请查看原文]
讨论
通过XRPD和拉曼光谱对犬结石的检查（图–）可以确定，其中存在的矿物形式由羟基磷灰石、镁方解石和无定形相组成。在经拉曼光谱分析的选择性富HA标本中，磷酸盐ν1谱带表明部分磷酸钙组分在结构上无序，并与ACP样贡献相容。虽然这些发现证实了HA是犬牙结石中主要的磷灰石形式，但镁方解石从未被提及作为牙结石的组分。这代表了对犬结石组成认识的重要进展，尽管Mg的浓度比Ca低近2个数量级（图）。在本犬样本中检测到的结晶磷酸钙相中羟基磷灰石占优势，可能反映了对相对成熟的块状沉积物的分析。这并不排除在结石矿化早期阶段其他亚稳态磷酸钙相的瞬态形成。
得益于考虑了足够大的宠物犬群体，我们证明了HA和CA相几乎以所有比例出现（图和），有助于解决关于犬结石组成的争论。事实上，文献中报道的相互矛盾的结果——一些声称方解石组成，一些声称磷灰石组成——源于分析样本组有限。
方解石和磷灰石组分的相对量取决于犬的个体特异性。为理解这一联系，需考虑最强的相关性评估于消融结石质量与距上次消融治疗的时间之间（图）。而HA含量则与消融质量（表）和犬年龄（表，图）显示明显相关性，同时与Ca%、Mg%、Ca/Mg、犬性别、品种和体重不相关。年龄–HA%数据的线性回归可估计HA增加速率约为∼4%/年，磷灰石含量的下限可能很难低于20%。
正如文献所述，宠物犬随着年龄增长自然处于更严重的牙周病阶段。根据我们的结果，由于牙周病与牙结石密切相关，结石质量的逐渐增加和HA组分随年龄的主导地位与口腔健康的恶化平行。HA组分增加的原因可能源于血清Ca2+水平（进而影响唾液中的Ca水平）与犬年龄之间已建立的负相关（独立于性别和体重）。因此，唾液中低Ca水平可能有利于HA的形成。然而，值得注意的是，牙结石中的Ca和Mg水平（图），连同Ca/Mg、Ca/P和P%，与犬年龄无显著相关性。注意Ca/P远高于纯HA的预期值（2.16），表明很大一部分Ca以非结晶含Ca有机来源组分的形式存在于结晶HA之外，并可能以ACC形式存在，ACC被公认为生物矿化中的主要前体相。
多变量分析阐明了单变量相关性的含义和局限性。HA富集分别与年龄、距上次牙齿清洁的时间以及清除的牙石质量相关，但这些变量在多元回归模型中并不表现为独立预测因子，这一事实表明犬DC组成更适合用成熟/积累耦合过程来描述，而非由单一控制因素决定。在此框架下，犬龄可能反映了与衰老相关的生物学和口腔环境变化，而清除质量和距上次清洁的时间则描述了沉积物的有效积累历史。PCA通过识别年龄、积累描述符和HA%共同变化的主轴支持这一解释（表S5和S6）。偏离该轴的样本在定义数据集变异性方面尤其具有信息价值。老年犬或较长积累时间后的富方解石沉积物，以及年轻犬的富磷灰石沉积物，可能代表对一般年龄-积累-组成关系的偏离。然而，由于本研究为横断面研究且缺乏标准化的牙周分期和无牙石对照，这些偏离应被视为未来纵向研究的假设而非诊断分类。
磷灰石含量与年龄的相关性强于与距上次清除时间的相关性（图S9），这表明DC组成不仅简单反映沉积物的时间生长时长，还受年龄相关生物学和口腔环境因素的影响。一致地，Ca/P比值与磷灰石分数倒数之间的关系（图）支持使用主量元素化学作为方解石-磷灰石平衡的独立替代指标。
[内容请查看原文]
结论
本研究表明，犬DC是一个化学和矿物学上异质的生物矿物系统，而非具有固定结晶组成的沉积物。XRPD分析确定羟基磷灰石和镁方解石为主要结晶相，存在于广泛的方解石-磷灰石连续体中。代表性样品的定量物相分析和选定富磷灰石标本的拉曼光谱进一步表明存在显著的非晶态贡献，包括与ACP相容的结构无序磷酸盐组分以及与ACC一致的碳酸盐贡献。
XRPD和ICP-OES的结合表明，XRPD衍生的结晶羟基磷灰石分数反映在沉积物的元素组成中。特别是，磷随羟基磷灰石富集单调增加，而钙和镁保持相对恒定。显著高的Ca/P比值及其随羟基磷灰石分数增加的非线性下降表明，相当一部分钙并非赋存于结晶羟基磷灰石中，支持含钙碳酸盐和非晶态组分的存在。因此，主量元素化学为犬DC中的方解石-磷灰石平衡提供了独立的描述符。
相关性和探索性多变量分析表明，羟基磷灰石富集与犬龄、清除的牙石质量和估计积累时间呈中度关联。这些变量不作为独立预测因子起作用；相反，它们沿年龄-积累-组成联合轴共同变化。观察到的从富方解石向富磷灰石沉积物的转变与矿物成熟过程一致，但研究的横断面设计不能确立方解石向磷灰石的时间性进展。
此处报道的矿物学趋势不应被解释为经过验证的诊断或预后工具。相反，它们为犬牙石建立了矿物学和化学框架，并为未来旨在解析位点特异性异质性、时间性矿物成熟以及牙石组成与犬口腔健康可能关系的对照研究提供了基础。尽管存在这些局限性，偏离预期矿物成熟趋势的情况，如年轻犬中晚期富磷灰石矿化或短时间间隔内不成比例的牙石积累，为未来纵向研究提供了可检验的假设，旨在确定偏离年龄-积累-组成关系是否与矿物成熟或口腔疾病进展的差异相关。
补充材料
[内容请查看原文]
支持信息可在https://pubs.acs.org/doi/10.1021/acsomega.6c07170免费获取。
为28只分析犬收集的信息（表S1）；定量XRPD物相分析（图S1、表S2a、S2b和S3）；全组样品的XRPD（图S2）。提取HA%的分布（图S3）；磷酸盐区域的拉曼光谱拟合，非晶/结晶和碳酸盐/磷酸盐比值的估计（图S4–S7、表S4）；清除牙石质量与HA%的线性拟合（图S8）；HA%与距上次牙齿清洁月数的线性拟合（图S9）；探索性多变量统计分析（多元线性回归模型和主成分分析）结果摘要（表S5、S6）（PDF）。
M.C. 构思了本研究，撰写了初稿，参与了数据获取与形式分析以及经费支持获取。A.L. 和 M.T. 进行了拉曼光谱分析并参与稿件编辑。C.P. 实施了消融、提供样本、收集犬只信息并参与稿件审阅。V.C. 参与经费获取、形式分析及稿件编辑/审阅。G.C. 参与数据获取、形式分析、经费获取及稿件编辑/审阅。
我们感谢米兰比可卡大学通过编号 2024-ATE-0525 的资助项目提供的经费支持。
使用 ChatGPT（OpenAI）协助统计报告。所有数值分析、解释和最终文本均由作者核查、严格修订并批准，作者对内容承担全部责任。
作者声明无竞争性经济利益。
参考文献
[内容请查看原文]
关联数据
补充材料
[内容请查看原文]
查看英文原文▾
Abstract
Dental calculus is a mineralized oral biofilm widely occurring in domestic dogs, yet its mineralogical variability and chemical evolution remain insufficiently defined. Here, we investigate dental calculus collected as biological waste material during routine professional dental cleaning from twenty-eight indoor pet dogs of different breeds, ages, and clinical backgrounds. X-ray powder diffraction, vibrational spectroscopy, and elemental analysis were combined to determine the crystalline phase assemblage, phosphate and carbonate spectral signatures, and major-element composition of the deposits. The crystalline fraction is dominated by hydroxyapatite and magnesian-calcite, which occur over a broad compositional range from calcite-rich to apatite-rich deposits. Raman analysis of selected apatite-rich samples further indicates the presence of structurally disordered phosphate contributions compatible with amorphous calcium phosphate, while quantitative phase analysis of a representative apatite-rich sample reveals a substantial amorphous fraction. Elemental data show that phosphorus increases monotonically with the crystalline hydroxyapatite fraction, whereas calcium and magnesium remain comparatively constant across the sample set. The Ca/P ratio is markedly higher than that of stoichiometric hydroxyapatite and decreases nonlinearly with increasing hydroxyapatite fraction, supporting the presence of calcium-bearing nonapatitic components and confirming major-element chemistry as an independent descriptor of the calcite–apatite balance. Correlation and exploratory multivariate analyses indicate that hydroxyapatite enrichment is moderately associated with dog age, ablated calculus mass, and estimated accumulation time, defining an age- and accumulation-associated mineralization continuum rather than a fixed calculus composition. These results provide a mineralogical and chemical framework for canine dental calculus as a heterogeneous biomineral system and suggest that its composition records coupled processes of biofilm mineralization, deposit accumulation, and mineral maturation. The study is cross-sectional and does not validate diagnostic use, but establishes a basis for future controlled and longitudinal investigations of dental calculus mineralization in canine oral health.
[内容请查看原文]
Introduction
Dental plaque is one of the most common and extensively investigated biofilms in animals. Its development begins with the adsorption of salivary proteins onto the tooth surface and the adhesion of pioneer microorganisms, followed by interbacterial coadhesion and microbial succession toward a more complex community. Plaque mineralization with formation of dental calculi (DC, also referred to as tartar), occurs when the local fluid becomes supersaturated with respect to calcium-phosphate salts, a condition strongly influenced by pH and alkalinizing bacterial metabolism. Mineral nucleation may occur heterogeneously on bacterial cell surfaces, membrane phospholipids and proteolipids, extracellular-matrix components, membrane-derived vesicles, and cellular debris. Salivary proteins, pyrophosphate, magnesium, and other promoters or inhibitors can further regulate nucleation, crystal growth, and phase selection. , Mechanical removal of dental biofilm is common practice in humans to prevent acids and toxins produced by biofilm bacteria from attacking enamel, damaging periodontal tissues, and inducing gingival inflammation, with eventual tooth loss. This prevents the formation of significant amounts of DC in humans, whereas, in pets, e.g., domestic dogs, regular dental cleaning is impractical and formation of considerable amounts of DC is rather common.
The global domestic canine population is continuously growing, reaching today an estimated number of half a billion, and the high prevalence of oral disorders in pet dogs has been well assessed in recent studies. , The most frequent oral disorder is periodontal disease (PD), with DC being an inherent component of the former. It is generally accepted that the inflammation and resulting tissue damage is due to an improperly regulated immune response to bacterial infection. , Given the potential link between DC and systemic health, scientific efforts should first address their physical-chemical characteristics, and then their relationship with canine health status.
Several calcium-phosphate phases have been identified in human DC, including brushite or dicalcium phosphate dihydrate, octacalcium phosphate, Mg-containing whitlockite, and carbonate-containing hydroxyapatite. Brushite is predominantly associated with early supragingival deposits, whereas octacalcium phosphate, hydroxyapatite, and whitlockite have been reported in both supragingival and subgingival calculus. Current models propose that calculus mineralization may initially involve amorphous or poorly ordered calcium phosphate and kinetically favored precursor phases such as brushite and octacalcium phosphate, followed by hydrolysis and maturation toward less soluble hydroxyapatite and whitlockite. This evolution should not be regarded as a unique linear pathway, because local pH, fluid composition, magnesium, fluoride, and macromolecular inhibitors can favor or stabilize different phases. ,
The most practical way to analyze DC is to determine their mineral composition. LeGeros & Shannon pioneered the investigation of the crystalline composition of DC in dogs, analyzing samples from approximately ten individuals. They illustrated by X-ray methods that the main components are Ca-carbonate in the form of calcite, and Ca-phosphate in the form of apatite. The calcite/apatite ratio was observed to vary from dog to dog because of variation of the carbonate/phosphate concentration in the dog saliva. This ratio was higher in the innermost layer of the calculus sample compared to the outermost layer, suggesting that calcite is prone to hydrolysis to apatite in the presence of phosphate in dog saliva. The predominance of calcite mixed with small amounts of apatite is a consequence of both the conditions of high pH and high carbonate concentrations in the dog saliva, together with the higher pH of resting plaque induced by the proteolytic activity. Indeed, the dog saliva is supersaturated with respect to calcium carbonate, and phosphate anions select the precipitation of Ca-carbonates in the form of the calcite polymorph.
Later, LeGeros et al. evaluated the efficacy of scanning electron microscopy (SEM) and electron probe microanalysis to reveal the solid components of human and animal DC, including dogs. In that study, quantitative analysis of Ca, Mg and the Ca/P ratio are reported, showing that the Ca/P molar ratio of animal calculi sets above 2.5 (1.3–1.7 in human) and the Mg content is not correlated with the Ca/P ratio. Ratios above 15 were attributed to regions exclusively composed by calcite.
Recently, Chiba-Ohkuma et al. provided a multimethodological analysis of seven, aged, indoor dogs (>15 y), confirming the higher Ca/P ratio in dog calculi with respect to human ones and also a larger fraction of organic matter, attributed to the absence of amylase in dog saliva. X-ray diffraction analysis suggested a mixture of calcite and carbonate-bearing hydroxyapatite.
Nakashima et al., starting from the evidence of ease of detachment of dog calculi as compared to human ones, analyzed the morphology and composition of the enamel/calculus interface. They showed the presence of a smooth surface and a 100 μm-wide gap demonstrating the poor adhesion. Microanalysis showed that the surface in contact with the enamel is characterized by a reduced content of Ca and P, which is not sufficient to establish if it is calcite-rich or phosphate-rich. In relation to the studies reported by Borah et al. −establishing that the inorganic components of canine saliva constrain the precipitation of carbonated hydroxyapatite −the authors claim that further studies are needed to bridge the mineralogical gap in canine calculus composition, in particular due to the pivotal role of Ca-salts for calculus formation.
The ionized form of Ca is the most important in biochemical reactions. More severe PD staging are generally associated with lower serum Ca2+ values. Interestingly, statistically significant correlations were observed between age, serum Ca2+ levels, and PD stage, while no such correlations were established with sex or weight. PD manifests approximately in 80–85% of dogs older than 2 y, with the disease affecting nearly all animals older than 5 y, with older dogs exhibiting more advanced stages of the disease, since PD stage becomes worse as the patient grows older.
In this work, we combine quantitative mineralogical analysis, major-element determination, and exploratory statistical analysis on DC collected from a population of 28 dogs. The results show that canine DC is not characterized by a fixed mineralogical composition, but spans a broad calcite–apatite continuum. This compositional variability defines a mineralogical maturation/accumulation trajectory, ranging from calcite-rich deposits, interpreted as comparatively early stage or less mature mineralization products, to apatite-rich deposits, representative of a more advanced mineralization state.
[内容请查看原文]
Materials and Methods
[内容请查看原文]
Sampling of Dental Calculi
DC were obtained as biological waste material during routine professional dental cleaning procedures using an ultrasonic scaler under general anesthesia performed by a licensed veterinarian (Clinica Veterinaria Bicocca, Milano, Italy) for oral hygiene and disease prevention purposes of 28 indoor dogs of various breeds. The collected material included both supragingival and subgingival calculus. Because routine ultrasonic scaling progressively detaches material from both locations, reliable separation of the two fractions during the clinical procedure was not practically feasible. They were therefore analyzed as bulk calculus. Accordingly, the present study addresses the overall mineralogical composition of the material removed during professional dental cleaning and does not resolve site-specific differences between supragingival and subgingival deposits. Dogs were included opportunistically when professional dental cleaning was clinically indicated by the veterinarian and when owner consent for anonymized scientific use of the removed material was obtained. No selection was made on the basis of breed, sex, reproductive status or expected mineralogical composition. No procedures beyond standard veterinary care were carried out for research purposes. The study protocol was evaluated by the Organismo Preposto al Benessere Animale (OPBA; Animal Welfare Body, AWB) of Clinica Veterinaria Bicocca (Milano, Italy), which confirmed that formal ethical approval was not required under Legislative Decree 26/2014 and Directive 2010/63/EU, as the research involved only biological waste material obtained during routine veterinary dental procedures and no experimental interventions were performed on live animals. The study is reported in accordance with the ARRIVE guidelines (https://arriveguidelines.org). No animals were euthanized for this study. Informed consent was obtained in written form from all dog owners prior to the clinical procedure and sample collection. Owners were informed that part of the material removed during professional cleaning could be anonymized and used for scientific analysis of mineral composition. All samples were handled following GDPR-compliant anonymization protocols to ensure that no identifying information about the dogs or owners could be retrieved.
Clinical information was retrieved from routine veterinary records. A standardized periodontal staging protocol including systematic periodontal probing and intraoral radiography was not performed for research purposes. Therefore, the oral-health score used here (see section ) should be considered an exploratory ordinal descriptor based on reported clinical observations, not a formal periodontal diagnosis. For each dog, the following information was recorded: ID, age, breed, weight, sex and reproductive status, history of previous dental cleanings, feeding habits, reported pathologies and surgical procedures, as well as the main outdoor environments frequented. Calculus samples were weighted (the mass of collected calculi samples varied from 10 to 800 mg) and dried in air for 24 h after extensive rinsing with deionized water. Because collection during routine ultrasonic scaling was not designed to quantitatively recover the entire calculus burden, this variable represents the mass of material collected for analysis rather than the total amount of calculus present in the oral cavity. It was therefore used as an operational proxy for calculus accumulation and should not be interpreted as an exact measure of total calculus burden. Sample codes (e.g., MNU, MLU, GMA, etc.) refer to anonymized individual dog identifiers. Collected information related to the investigated dogs is arranged in Table S1.
Samples were powdered in a mortar and stored in desiccators. Subsequent analyses were performed using the following methods.
[内容请查看原文]
X-Ray Powder Diffraction (XRPD)
An X’Pert Pro X-ray diffractometer (Malvern Panalytical) with Bragg–Brentano geometry (specular scans) and Cu target (λ = 1.54 Å) as X-ray source was used to collect diffraction patterns on the powdered samples to determine mineral composition and relative occurrence. Given the low amount of sample available for some dogs and to adopt a common methodological approach for all of them allowing for a consistent comparison, data collection was performed using a low-background planar sample holder, enabling the measurement of very small sample amounts. However, while delivering a data set suitable for reliable determination of calcite and apatite proportions, this technique could not be employed to determine the amorphous content accompanying the crystalline phases, which requires the mixing with a known amount of an internal standard (corundum), making this mixed sample no longer suitable for subsequent analyses. On the contrary, for a representative voluminous samples (LLI, mass = 320 mg), data collection was also performed with corundum internal standard, enabling quantitative phase analysis (QPA) also of the amorphous fraction (see Supporting Information). QPA was undertaken by Rietveld refinement using Profex v. 5.6.1.
[内容请查看原文]
Raman Spectroscopy
The micro-Raman measurements were carried out with the HR800 Horiba Jobyn-Yvon instrument using a 50× objective (0.75 NA, laser spot size about 1 μm) and the 405 nm excitation provided by a solid-state laser produced by Integrated Optics. Before measurements, the sample powder was gently pressed on a microscope glass slide. For each different point on the powder sample, the laser focusing procedure was repeated before collecting the Raman spectra. The different points on the sample were randomly selected within a zone of a few millimeters in diameter. For each powder sample, we collected several spectra by focusing the laser on different points of the available material. We observed negligible variations from point to point; thus, the reported spectra can be considered representative of each sample. The laser power incident on the sample was kept below 10 mW for all measurements to avoid photodamage.
[内容请查看原文]
Quantitative Analysis of Major Elements
Powder samples were dissolved in a mixture of 4 mL HNO3 65%, 4 mL HCl 37%, and 2 mL H2O2 40% by the aid of a microwave digestor (MV 5000, Anton Paar) undertaking a thermal treatment at 200 °C for 10 min. The concentration of Ca, Mg, and P contained in the dissolved calculi were determined by using an Inductively Coupled Plasma-Optical Emission Spectroscopy (ICP-OES) Optima 7000 DV (PerkinElmer, Italy) system.
[内容请查看原文]
Statistical Analysis
A unified data set was constructed by matching each DC sample with the corresponding dog record using the anonymized sample ID. The mineralogical descriptor used for statistical analysis was the hydroxyapatite fraction obtained by Rietveld refinement with Profex, hereafter indicated as HA%. The calcite fraction was calculated as CA% = 100 – HA%, considering only the crystalline fraction resolved by XRPD. Additional quantitative variables included dog age at ablation, body weight, ablated calculus mass, time elapsed since the previous dental cleaning, and the elemental concentrations of Ca, Mg and P determined by ICP-OES. The Ca/Mg and Ca/P weight ratios were calculated from the corresponding elemental concentrations.
The time elapsed since the previous dental cleaning was defined as the number of months between birth and the sampling ablation for dogs undergoing their first dental cleaning, and as the number of months between the previous dental cleaning and the sampling ablation for dogs that had undergone one or more previous dental cleaning procedures. This variable was therefore used as an estimate of the effective accumulation time of the calculus deposit.
Reported oral conditions were converted into an ordinal oral-health score for exploratory statistical analysis. Dogs with no reported systemic or oral pathology were assigned score 0. Dogs with gingivitis, including minor, modest or moderate gingivitis, were assigned score 1. Dogs with modest or moderate periodontitis were assigned score 2. Dogs with periodontitis were assigned score 3. Dogs with severe periodontitis and/or tooth mobility were assigned score 4. Dogs for which oral-health information was not available were excluded from analyses involving the oral-health score. Systemic conditions, when reported separately from oral conditions, were not included in the oral-health score. Since this score is not a formally validated periodontal stage, its relationship with other parameters (e.g., the fraction of mineral phases) has no predictive significance.
Univariate associations between quantitative variables were evaluated using both Pearson’s correlation coefficient (r), to assess linear relationships, and Spearman’s rank correlation coefficient (ρ), to assess monotonic relationships less sensitive to nonlinearity and outliers. Simple linear regressions were performed for selected relationships, including HA% as a function of dog age, HA% as a function of ablated calculus mass, ablated calculus mass as a function of time elapsed since the previous dental cleaning, and Ca/P as a function of HA% or its reciprocal transformation. The reciprocal transformation 100/HA% was also considered because the Ca/P ratio is expected to vary nonlinearly with the relative abundance of the phosphate-bearing crystalline phase.
Exploratory multivariate analysis was performed using Python 3 with the pandas, scipy, statsmodels and scikit-learn packages, to evaluate whether the compositional variability of canine DC could be described by coupled age-, accumulation- and composition-related descriptors. Multiple linear regression models were fitted using statsmodels, with HA% as the dependent variable and dog age, time elapsed since the previous dental cleaning, ablated calculus mass and, where appropriate, oral-health score as explanatory variables. Variance inflation factors were calculated to evaluate collinearity among predictors.
Principal component analysis was carried out on standardized variables to identify latent axes of covariation within the data set. A first principal component analysis (PCA) was performed using scikit-learn after standardization of the input variables on the core quantitative variables available for all dogs, namely dog age, time elapsed since the previous dental cleaning, ablated calculus mass and HA%. A second PCA was performed on the subset of samples for which complete chemical data were available, including HA%, 100/HA%, dog age, time elapsed since the previous dental cleaning, ablated calculus mass, Ca%, Mg%, P% and Ca/P. PCA was used as an exploratory tool and not as a predictive model. Scores and loadings were inspected to identify compositional trends, accumulation-related trends and samples deviating from the main age–accumulation–composition axis.
All statistical analyses were performed on the complete 28-dog data set when the variables were available. Analyses involving oral-health score were restricted to dogs with known oral-health information. Statistical significance was assessed at p < 0.05. Given the limited cohort size and the exploratory nature of the multivariate analysis, results were interpreted primarily in terms of effect size, direction of association and consistency among variables rather than as definitive causal evidence.
[内容请查看原文]
Results
[内容请查看原文]
Mineralogical Characterization of Dental Calculi
An accurate quantitative XRPD measurement performed on a representative and particularly massive sample (LLI) is reported in Figure . The full indexation of the diffraction peaks is accomplished by considering the presence of hydroxyapatite (HA, Ca5(PO4)3(OH)) and magnesian calcite [CA, (Ca,Mg)CO3]. Related peak positions, hkl indexes and relative intensities of major peaks are reported in Table S2a. The relatively broad peaks of the HA phase indicate its lower degree of crystallinity as compared to CA. Accordingly, besides the two minerals HA and CA, an amorphous phase composes the sample for a fraction as high as 61.4% (Figure S1 and Table S2b).
[图：Sample LLI. Experimental (black) vs calculated (red) diffraction pattern and related intensity difference (gray). Major ]
Magnesian calcite was suggested as the highest matching pattern in the 87% of cases (see Supporting Information). The amount of MgCO3 estimated from the refined a parameter is between 0.86 and 2.33 mol %, with an average of 1.16 mol % (Table S3). On this basis, the mineral can be defined a low-Mg calcite.
After establishing that these three phases compose the canine calculi, their relative amount can be estimated by comparing the intensity of the corresponding XRPD reflections via Rietveld refinement.
Figure reports a selection of XRPD patterns – only every other pattern along the series reported to improve readability. The whole series of 28 XRPD patterns acquired on an equal number of different samples is reported in the Supporting Information as Figure S2. As can be observed, the amount of HA and CA spans all the proportions, from almost purely apatitic, to almost purely calcitic (Figure ), with a distribution of the HA% measured showing an increasing trend (Figure S3). The visual inspection of Figure confirms the reliability of QPA. Actually, the main diagnostic peaks of the two crystalline phases vary consistently along the series, with few exceptions, i.e., PTA, which probably should occupy a lower position (see also PTO in Figure S2). However, it should be noted that both PTA and PTO belong to the HA-richest region characterized by a plateau in Figure , where the relative proportions between HA and CA vary within the experimental error (Table S3).
[图：Selected XRPD patterns of DC – only every other pattern along the series has been selected – showing a continuous variat]
[图：Variation of the HA% and CA% and estimated standard deviations within the crystalline fraction of the complete sample se]
Raman spectroscopy was used as a complementary phase-sensitive probe to support the identification of carbonate- and phosphate-bearing components and, through analysis of the phosphate ν1 band, to assess the structural order of the calcium-phosphate fraction. The results obtained for the LLI sample are reported in Figure , covering the spectral range from 200 to 2000 cm–1. The most intense peaks, clearly emerging from a broad fluorescence background, are assigned to phosphate (∼960 cm–1) and carbonate (∼1070 cm–1). Additionally, less intense features associated with the bending modes of phosphate around 400 and 600 cm–1 can be observed, , along with two weaker bands (CH2 wagging and Amide I) attributed to proteins in hydroxyapatite of biogenic origin. Additionally, a weak peak at 1004 cm–1 can be tentatively attributed to phenylalanine.
[图：Raman spectrum of the LLI sample (405 nm excitation, ca. 10 mW power at the sample, 50× objective). Black arrows indicat]
On a closer look, the phosphate band appears structured and displays two main components, one at 950 cm–1, assigned to the amorphous fraction of the sample, and another at 960 cm–1, assigned to the crystalline part (Figure S4). Indeed, the HA crystalline phase often derives from a less ordered, metastable precursor known as amorphous calcium phosphate (ACP, Ca x (PO4) y ·nH2O; n = 3–4.5). The relative amount of ACP and HA can be effectively monitored by Raman spectroscopy, since the phosphate symmetric stretching band shifts from approximately 950 cm–1 (amorphous phase) to 960 cm–1 (crystalline hydroxyapatite), accompanied by a progressive narrowing of the peak. This diagnostic behavior is described in ref and provides a reliable spectroscopic approach to characterize the amorphous and crystalline phosphate fractions. For each spectrum, the region of the phosphate band (900–1020 cm–1) was selected, and the signal was deconvoluted into two Lorentzian components centered at ∼950 cm–1 and ∼960 cm–1, corresponding respectively to the amorphous (A) and crystalline (C) phosphate contributions. The linear baseline was subtracted, and fits were performed with R 2 > 0.98, following the procedure outlined in ref . The ratio of the areas of the amorphous and crystalline components (A/C = A950/A960) was then used as a semiquantitative indicator of the relative amorphous content in each specimen. The individual fits are shown in Figure S4 for the HA-rich samples LLI, LPE, SIA and MRA, and the results of the fitting for all spectra are summarized in Table S4. The distribution of the A/C ratios among the four samples is illustrated in Figure S5 as a bar chart. The data indicates some differences in phosphate ordering; for instance, the LLI sample exhibits a more disordered, less crystalline phosphate structure, as compared with the SIA and LPE samples. In the LLI sample, the Raman analysis allows us to conclude that, within the total amount of Ca-phosphate, 31% is constituted by ACP. Since the weight fraction of crystalline HA estimated in LLI is 37% (Table S2b), the weight fraction of ACP is deduced to be 17%. The remaining amorphous fraction – the overall amount estimated by Rietveld refinement is ∼61.4% – comes from other phases, probably of biogenic origin (vide infra). The ACP fractions in Ca-phosphate of samples LPE, SIA and MRA are 17%, 20% and 25%, respectively (Table S4).
As shown in the Figure S6, the carbonate band at ∼1070 cm–1 actually comprises two distinct contributions: one at ∼1072 cm–1 associated with carbonate ions substitutional to phosphate (B-type) within the hydroxyapatite lattice or to amorphous calcium carbonate (ACC), , and another from carbonate ions associated with magnesian calcite (∼1086 cm–1). To investigate the possibility of qualitatively assessing the variations of the relative amounts of all carbonate and all phosphate groups by Raman, we collected spectra in the 900–1200 cm–1 region for the LLI, SIA, LPE, and MRA samples. We performed spectral fitting over a linear baseline to evaluate the area of the carbonate band (∼1070 cm–1) and compare it (carbonate/phosphate ratio) to that of the phosphate band (∼960 cm–1). The spectral fits are presented in Figure S7a, whereas the carbonate-to-phosphate band area ratio is shown in Figure S7b. We observe that the carbonate/phosphate ratio monotonically decreases from 0.45 to 0.20 along the sequence LLI, SIA, LPE, MRA (Figure S7b), whereas the CA/HA ratio obtained from XRD data is virtually constant (Figure ). However, it is essential to emphasize the fundamental difference between the two ratios obtained using these two techniques: in XRD, the ratio represents the relative proportions of the CA and HA crystalline phases, whereas in Raman spectroscopy, it reflects the relative amounts of carbonate and phosphate anions.
[内容请查看原文]
Assessment of Correlations between Mineralogical and Elemental Composition of Dental Calculi – Monovariate Statistical Analysis
Some correlations can be expected to link the mineralogical composition (relative fraction of HA and CA) to the elemental composition. Assuming a composition of exclusively CA and HA, with Ca contributing to 40%w/w both in CA and in HA, and with P contributing to 18.5%w/w in HA, these relationships are expected:
CaP=2.16(100HA%);P%=0.185HA%The content of major elements and Mg of the ablated calculi is reported as weight percent as a function of apatite content in Figure . Calcium is the most abundant element with a fraction of 28% ± 9%, whereas the Mg content sets to 0.34% ± 0.12%. It is worth noting that Ca% and Mg% are not correlated with the apatite content, whereas P% augments monotonically but not linearly (Pearson’s r = 0; Spearman ρ = 0.536, p = 0.0122, Table ) from 0.5% to 1.5% with the apatite content.
[图：Weight percent of Ca (red dots, left axis), Mg (green dots, right axis), and P (blue dots, right axis) in the analyzed c]
1. Estimated Correlations between HA% and Elemental Composition of Canine Calculi
| Relation | test | coefficient | p -value | comment |
|---|---|---|---|---|
| P% vs HA% | Spearman | ρ = 0.536 | p = 0.0122 | significant, monotone |
| Ca/P vs 1/HA% | Pearson | r = 0.713 | p = 0.000285 | Strong and significant |
Figure a shows the fractions Ca/Mg and Ca/P as a function of the apatite content. While the ratio Ca/Mg is not correlated to HA% (it sets to 87 ± 26), the Ca/P ratio shows a decreasing trend with HA%, with a strong linear correlation with 1/HA% (Figure b; r = 0.713, p = 0.000285, Table ). Considering that the Ca/P weight ratio and P% expected for pure HA are 2.16 and 18.5, respectively, while the corresponding estimated experimental values are 13 and 1.5, a significant portion of Ca must be hosted outside crystalline HA, including the crystalline carbonate phase identified by XRPD and potentially noncrystalline Ca-bearing components. On the light of the strong band at 1072 cm–1 observed in Raman spectra (Figures and S6), this fraction of amorphous phase may include ACC.
[图：(a) Weight ratios of Ca/Mg (black circles) and Ca/P (red circles) in the analyzed calculi as a function of their HA cont]
[内容请查看原文]
Assessment of Correlations between Calculus and Dog Specificities – Monovariate Statistical Analysis
The strongest correlation is observed between ablated mass and age/months from last ablation (Figure ).
[图：Ablated calculus mass as a function of the (a) dog age and (b) months from last ablation. The solid red lines represent ]
The ablated calculus mass correlates more strongly with the time elapsed since the previous dental cleaning than with chronological age, indicating that the amount of collected calculi is associated with the estimated accumulation time of the deposit.
In relation to possible correlations between mineral composition and the specificity of the dogs, a significant but moderate correlation is also assessed between HA% and the ablated calculus mass (Table and Figure S8), time elapsed from a previous ablation (if present, Figure S9) and dog age (Table and Figure ). Variable that does not show significant correlation with HA% (together with Ca% and Mg% and Ca/Mg mentioned in section ) is dog weight.
2. Estimated Correlations between HA% and Some Dog Specificities
| Relation | test | coefficient | p -value | comment |
|---|---|---|---|---|
| Ablated mass vs HA% | Pearson | r = 0.497 | p = 0.0071 | significant |
| Spearman | ρ = 0.548 | p = 0.0025 | significant | |
| Months from last ablation vs HA% | Pearson | r = 0.407 | p = 0.0315 | significant |
| Spearman | ρ = 0.455 | p = 0.0150 | significant | |
| Age vs HA% | Pearson | r = 0.455 | p = 0.0149 | significant |
| Spearman | ρ = 0.403 | p = 0.0333 | significant |
[图：Weight percent of HA as a function of the dog age. The thick red dashed line represents the linear regression line of th]
Principal outliers in respect to the HA%-age relationship are five dogs for their anomalously high CA% in their calculi (AGY, TEO, SIA2, WSP, and HIE), and one dog for its anomalously high HA% (PTA). The calcite-rich negative outliers are characterized by relatively low ablated calculus masses (Table ) and by mild-to-moderate oral diseases (Figure ).
3. Distribution of Ablated Calculus Mass among the Considered 28 Dogs Divided in a Group of Calcitic Outliers (See Figure ) and All Other Dogs
| Group | n | Average mass (mg) | Median mass (mg) | Range (mg) |
|---|---|---|---|---|
| Calcitic outliers | 5 | 110 | 110 | 30–170 |
| others | 23 | 277 | 240 | 10–800 |
To increase the group of positive outliers and balance the comparison, next to PTA we can include AES, MRA and SIA (Figure ). These outliers are associated with relatively high ablated masses of 160, 160, and 800 mg, respectively, and their oral health is associated with severe periodontitis or mild periodontitis (Figure ). Among the outliers of Figure , only AES appears as outlier in Figure b as well. This means that strong deviations of the HA content in relation to age are not associated with anomalous Ca/P ratios.
Since HA% also increases with ablated mass (Figure S8, Table ), the apparent age dependence of the HA enrichment (Figure ) should be interpreted as reflecting a combined effect of dog age, calculus accumulation history, and mineral maturation.
[内容请查看原文]
Assessment of Correlations between Calculus and Dog Specificities – Multivariate Statistical Analysis
Exploratory multivariate analysis further supported the existence of a coupled age–accumulation–composition trend. A multiple linear regression model using dog age, time elapsed since the previous dental cleaning and ablated calculus mass as predictors of the XRPD-derived crystalline HA fraction explained approximately 29% of the variance in HA% (R 2 = 0.287; adjusted R 2 = 0.198; model p = 0.0405). However, none of the individual predictors retained statistical significance when considered simultaneously (Table S5), indicating that age, accumulation time and ablated mass are partially interdependent descriptors rather than independent determinants of the mineral composition. Variance inflation factors remained moderate, supporting the use of the model as an exploratory description of covariation. When the ordinal oral-health score was added to the model in the subset of dogs with available oral information, the explained variance increased only slightly and the model was not statistically significant, confirming that the available clinical descriptors should be interpreted cautiously and not as formal periodontal staging.
Principal component analysis of the four core quantitative variables available for all dogs (age, time elapsed since the previous dental cleaning, ablated mass and HA%) identified a dominant first component explaining 68.4% of the total variance. This component showed positive loadings for all four variables, consistent with a common age–accumulation–mineral maturation axis (Table S6). A second component, explaining 17.0% of the variance, mainly separated samples with relatively high HA% from those with comparatively higher age or accumulation time but lower HA%. This second axis therefore highlights samples deviating from the main maturation trend, including calcite-rich samples with relatively low HA% for their age or accumulation history and apatite-rich samples with comparatively high HA% despite limited accumulation-related descriptors. A second PCA performed on the subset with complete chemical data showed that HA%, 100/HA%, Ca/P and P% contributed strongly to the compositional variance, consistent with major-element chemistry providing an independent descriptor of the calcite–apatite balance.
[内容请查看原文]
Discussion
The examination by XRPD and Raman spectroscopy of canine calculi (Figures –) allows to establish that the mineral forms present therein are constituted by hydroxyapatite, Mg-calcite, and an amorphous phase. In the selected HA-rich specimens analyzed by Raman spectroscopy, the phosphate ν1 band indicates that part of the calcium phosphate component is structurally disordered and compatible with ACP-like contributions. While these findings corroborate HA as the dominant apatitic forms in dog DC, , Mg-calcite was never mentioned as DC component. This represents an important increment of the knowledge of the composition of dog calculi, even though Mg occurs with concentrations almost 2 orders of magnitude lower than Ca (Figure ). The predominance of hydroxyapatite among the crystalline calcium-phosphate phases detected in the present canine samples may reflect the analysis of relatively mature bulk deposits. It does not exclude the transient formation of other metastable calcium-phosphate phases during earlier stages of calculus mineralization.
Thanks to the consideration of a sufficiently large population of pet dogs, we demonstrated that HA and CA phases occur virtually in all proportions (Figures and ), contributing in solving the debate about the composition of canine calculi. Indeed, the contrasting results reported in the literature, with some claiming a calcitic composition and some claiming an apatitic composition, derives from the limited group of analyzed samples.
The relative amount of the calcitic and apatitic components depends on the specificity of the dog. To understand this link, consider that the strongest correlation is assessed between the mass of ablated calculi and the time elapsed from previous ablation treatments (Figure ). The HA content, in turn, shows a clear correlation with the ablated mass (Table ) and dog age (Table , Figure ), while it is uncorrelated with Ca%, Mg%, Ca/Mg, dog sex, breed and weight. The linear regression of the age-HA% data allows to estimate the rate of HA increment as ∼4% y–1, with a lower limit of apatite content which probably hardly sets below 20%.
As the literature illustrates, pet dogs are naturally subjected to more severe PD stages as their age increases. In the light of our results, being PD closely related to DC, the progressive increment of calculus mass and dominance of the HA component with age goes in parallel with a worsening of the oral health. The motivations behind the increment of HA component can originate from the established negative correlation between serum Ca2+ levels (affecting in turn Ca levels in saliva) and dog age (independently of sex and weight). Hence, low Ca levels in saliva may favor the formation of HA. However, it is worth observing that Ca and Mg levels in DC (Figure ), together with the Ca/Mg, Ca/P, and P%, display no significant correlation with dog age. Note that Ca/P is much higher than that expected for pure HA (2.16), indicating that a substantial fraction of Ca is hosted outside crystalline HA as noncrystalline Ca-bearing components of organic origin and possibly in the form of ACC, which is well recognized as a primary precursor phase in biomineralizations.
The multivariate analysis clarifies the meaning and the limitations of the univariate correlations. The fact that HA enrichment correlates individually with age, time elapsed since the previous dental cleaning and ablated calculus mass, but that these variables do not behave as independent predictors in the multiple regression model, indicates that canine DC composition is better described by a coupled maturation/accumulation process than by a single controlling factor. In this framework, dog age likely captures biological and oral-environmental changes associated with aging, whereas ablated mass and time since previous cleaning describe the effective accumulation history of the deposit. PCA supports this interpretation by identifying a dominant axis along which age, accumulation descriptors and HA% covary (Tables S5 and S6). Samples deviating from this axis are particularly informative in defining the variability of the data set. Calcite-rich deposits in older dogs or after longer accumulation times, as well as apatite-rich deposits in younger dogs, may represent deviations from the general age-accumulation-composition relationship. However, because the study is cross-sectional and lacks standardized periodontal staging and calculus-free controls, these deviations should be regarded as hypotheses for future longitudinal studies rather than as diagnostic classifications.
The stronger correlation of apatite content with age than with time elapsed since the last ablation (Figure S9) suggests that DC composition does not simply reflect the chronological growth time of the deposit, but is also influenced by age-related biological and oral-environmental factors. Consistently, the relationship between the Ca/P ratio and the inverse apatite fraction (Figure ) supports the use of major-element chemistry as an independent proxy for the calcite–apatite balance.
[内容请查看原文]
Conclusions
This study shows that canine DC is a chemically and mineralogically heterogeneous biomineral system rather than a deposit with a fixed crystalline composition. XRPD analysis identifies hydroxyapatite and magnesium-calcite as the main crystalline phases, occurring across a broad calcite–apatite continuum. Quantitative phase analysis of a representative sample and Raman spectroscopy of selected apatite-rich specimens further indicate that a substantial amorphous contribution is present, including structurally disordered phosphate components compatible with ACP and carbonate contributions consistent with ACC.
The combination of XRPD and ICP-OES demonstrates that the XRPD-derived crystalline hydroxyapatite fraction is reflected in the elemental composition of the deposits. In particular, phosphorus increases monotonically with hydroxyapatite enrichment, whereas calcium and magnesium remain comparatively constant. The markedly high Ca/P ratios, together with their nonlinear decrease with increasing hydroxyapatite fraction, indicate that a significant fraction of calcium is not hosted by crystalline hydroxyapatite, supporting the presence of calcium-bearing carbonate and amorphous components. Major-element chemistry therefore provides an independent descriptor of the calcite–apatite balance in canine DC.
Correlation and exploratory multivariate analyses indicate that hydroxyapatite enrichment is moderately associated with dog age, ablated calculus mass, and estimated accumulation time. These variables do not act as independent predictors; rather, they covary along a combined age-accumulation-composition axis. The observed shift from calcite-rich to apatite-rich deposits is consistent with a mineral-maturation process, but the cross-sectional design of the study does not establish a temporal calcite-to-apatite progression.
The mineralogical trends reported here should not be interpreted as a validated diagnostic or prognostic tool. Rather, they establish a mineralogical and chemical framework for canine dental calculus and provide a basis for future controlled studies aimed at resolving site-specific heterogeneity, temporal mineral maturation, and the possible relationship between calculus composition and canine oral health. Notwithstanding these limitations, deviations from the expected mineralogical maturation trend, such as advanced apatite-rich mineralization in young dogs or disproportionate calculus accumulation over short time intervals, provide testable hypotheses for future longitudinal studies aimed at determining whether departures from the age-accumulation-composition relationship are associated with differences in mineral maturation or oral-disease progression.
Supplementary Material
[内容请查看原文]
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c07170.
Information collected for the 28 analyzed dogs (Table S1); quantitative XRPD phase analysis (Figure S1, Tables S2a, S2b and S3); XRPD of the whole set of samples (Figure S2). Distribution of extracted HA% (Figure S3); raman spectral fitting in the phosphate region, estimation of the amorphous/crystalline and carbonate/phosphate ratios (Figures S4–S7, Table S4); linear fit of ablated calculus mass vs HA% (Figure S8); linear fit of HA% vs months from previous dental cleaning (Figure S9); summary of the results of the exploratory multivariate statistical analysis (multiple linear regression models and principal component analysis) (Tables S5,S6) (PDF)
M.C. conceptualized the study, wrote the original draft, contributed to data acquisition and formal analysis and to the acquisition of financial support. A.L. and M.T. performed the Raman spectroscopic analysis and contributed to editing the manuscript. C.P. performed ablation, provided samples, collected dog information, and contributed to reviewing of the manuscript. V.C. contributed to funding acquisition, formal analysis and editing/reviewing of the manuscript. G.C. contributed to data acquisition, formal analysis, funding acquisition and editing/reviewing of the manuscript.
We acknowledge funding by the University of Milano – Bicocca through grant n. 2024-ATE-0525.
ChatGPT (OpenAI) was used to assist statistical reporting. All numerical analyses, interpretations and final text were checked, critically revised and approved by the authors, who take full responsibility for the content.
The authors declare no competing financial interest.
References
[内容请查看原文]
Associated Data
Supplementary Materials
[内容请查看原文]