兽研血清蛋白毛细管电泳可将接受犬利什曼病（CanL）检测的犬的异常蛋白血症表征为体液免疫激活的综合临床病理表型。这项基于实验室的回顾性研究评估了2009年至2026年间在葡萄牙接受CanL检测的5544只犬的血清蛋白电泳模式。抗利什曼原虫抗体结果被分为五个有序类别：阴性、不确定、低阳性、阳性/高阳性和极高阳性。分析了总蛋白、白蛋白、α-1、α-2、β和γ球蛋白组分、白蛋白与球蛋白比值以及白蛋白与γ球蛋白比值。另定义一个由254只年龄≥1至<4岁的临床健康血清阴性犬组成的参考亚群，用于建立研究特定的非参数95%参考区间。排除不确定结果后，二分类分析纳入2454只血清阴性犬和2910只血清阳性犬。血清阳性犬的γ球蛋白中位数高于血清阴性犬（1.5 vs. 0.8 g/dL；p < 0.001），白蛋白与球蛋白比值（0.7 vs. 1.0；p < 0.001）及白蛋白与γ球蛋白比值（1.9 vs. 4.2；p < 0.001）较低。在有序血清学分级中，γ球蛋白从0.8升高至2.1 g/dL（p < 0.001），而白蛋白与γ球蛋白比值从4.2下降至1.2。γ球蛋白显示出最强的有序关联（ρ = 0.522；p < 0.001）以及对血清阳性的中等单标志物判别能力（受试者工作特征曲线下面积 = 0.768；95%置信区间：0.755–0.780）。这些发现将CanL蛋白图谱定义为一种进行性、可量化的免疫病理表型，以γ区带扩张和相伴随的白蛋白下降为主导，支持血清蛋白电泳作为解释和监测的参考工具。
1. 引言
血清蛋白电泳是少数能够将简单的生化异常转化为宿主免疫生物学可解释表征的常规实验室技术之一[1,2]。它不仅确认高蛋白血症或高球蛋白血症，还将血清蛋白组分解析为白蛋白、α、β和γ组分的结构化图谱，从而提供关于炎症、急性期激活、抗原刺激、免疫球蛋白产生、蛋白丢失和慢性组织损伤的见解[3,4,5,6]。
在犬中，血清蛋白电泳已从主要用于描述异常蛋白血症的技术，发展为识别炎症性、感染性、免疫介导性和肿瘤性模式的临床信息工具[4,5,6]。毛细管区带电泳因其自动化、分析速度快以及相比传统电泳方法分辨率提高而特别有价值[4,7]。然而，电泳解释仍具有平台特异性，结果不应在分析系统或实验室之间不加批判地转移[4,5]。绝对组分浓度由电泳百分比和血清总蛋白浓度计算得出，使该技术兼具定量和形态学信息价值[4,5]。然而，其临床价值取决于对完整电泳图谱的解释，而非孤立的组分数值。
每个电泳组分的生物学意义各不相同但又相互关联。白蛋白反映肝脏合成、胶体渗透压状态、负急性期行为和蛋白丢失。α组分包括多种急性期蛋白，因此常反映早期或活动性炎症[8,9]。β组分由一组异质性蛋白组成，包括转铁蛋白、补体、脂蛋白和某些免疫球蛋白，在炎症性、寄生虫性和免疫复合物相关疾病中可能尤为重要。在严重肾小球蛋白丢失的犬中，α-2和β区带也可能出现相对突出，因为α-2-巨球蛋白和脂蛋白等高分子量蛋白相对被保留。因此，这些变化应结合绝对组分浓度和肾脏参数进行解释。γ区带主要由免疫球蛋白组成，尤其是免疫球蛋白G（IgG），因此与慢性抗原刺激和体液免疫激活密切相关[3,4,5,6]。因此，蛋白图谱提供了炎症和免疫反应的系统层面视图，整合了通常通过生化、炎症标志物、血清学和临床分期分别评估的生物学过程[8,9,10,11,12,13]。
图1提供了正常犬血清蛋白图谱的示意图，以促进对本研究中讨论的电泳组分的解释。
犬利什曼病（CanL）主要由利什曼原虫（L. infantum）引起，是兽医学中研究慢性炎症和持续免疫刺激相关血清蛋白电泳改变的最佳自然发病疾病模型之一[14,15]。CanL是一种人畜共患的媒介传播寄生虫病，临床表现谱广泛，可从亚临床感染到严重的慢性全身性疾病[16,17,18,19,20]。犬是利什曼原虫（L. infantum）的主要家养储存宿主，疾病表现强烈依赖于寄生虫与宿主免疫应答之间的相互作用[16,21,22]。保护性细胞免疫可能有助于控制感染，而过度且持续的体液免疫激活则与疾病进展、循环免疫复合物、肾小球损伤、全身性炎症及特征性异常蛋白血症相关[16,17,23,24]。CanL中的电泳图谱并非被动的实验室异常，而是宿主免疫病理应答的生化表达。
与临床发病的利什曼病犬相关的经典血清蛋白电泳模式包括低白蛋白血症、白蛋白与球蛋白比值降低、多克隆高丙种球蛋白血症，以及在许多犬中出现的不同程度的β球蛋白增多和急性期改变[16,17,23,25,26]。这些发现从生物学上与慢性抗原暴露、非保护性抗体产生、免疫复合物形成、肾脏蛋白丢失及血清蛋白的炎症性再分布相一致[27,28]。这些异常均非利什曼病的特征性表现。定量血清学并非在所有抗体水平上都仅仅是暴露标志物：对于具有相符临床或临床病理学异常的犬，高抗利什曼原虫抗体滴度可确诊CanL，而低滴度则需进一步检查[16,17,29,30]。因此，SPE不应取代定量血清学或寄生虫检测；其原理在于将异常蛋白血症定位至白蛋白、急性期、β或γ区域，并显示宽基底、限制性或类单克隆形态，从而提供总蛋白或计算球蛋白 alone 无法获得的信息[3,4,5,6]。
血清蛋白电泳在犬媒介传播疾病中的相关性近期重新受到关注。在自然感染犬恶丝虫的犬中，电泳评估显示白蛋白和β球蛋白异常频繁出现，且治疗后白蛋白、β组分及白蛋白与球蛋白比值有所改善，支持蛋白图谱作为寄生虫病辅助监测工具的作用[31]。更直接地，在CanL中，一项前瞻性治疗监测研究报告，接受锑酸葡甲胺加别嘌醇治疗的犬在短期内白蛋白浓度和白蛋白与球蛋白比值升高，同时γ球蛋白降低，支持血清蛋白电泳作为早期评估治疗反应的补充工具[10]。此外，一项近期研究评估了对多种媒介传播病原体血清学阳性的犬的血清蛋白电泳和免疫固定电泳，显示大多数异常蛋白血症本质为炎症性或多克隆性，仅少数病例符合提示可能存在M蛋白的电泳和免疫固定电泳联合标准[32]。这些观察强调了两个重要原则：第一，某些媒介传播感染可诱导反映宿主免疫病理应答的特征性血清蛋白电泳模式；第二，对宽基底、限制性或类单克隆条带的解读需要生物学背景和分析上的谨慎[6,32,33]。
然而，在CanL中，蛋白图谱常被视为诊断的描述性附属内容，而非主要研究对象。许多研究报告了高丙种球蛋白血症、低白蛋白血症和白蛋白与球蛋白比值降低，但相对较少研究探讨电泳表型是否随抗利什曼原虫血清学强度增加而发生定量变化[14]。这一区别至关重要。血清学阴性与阳性犬之间的二元比较可能证实异常蛋白血症在感染中常见，但无法确定血清蛋白谱是否随体液免疫应答增强而逐步重塑。
此前在葡萄牙开展的一项大规模回顾性研究显示，在因利什曼原虫感染接受检测的犬中，γ球蛋白浓度和白蛋白与球蛋白比值是与抗利什曼原虫酶联免疫吸附试验（ELISA）分类最明显相关的临床病理学变量之一[14]。该研究提供了重要的流行病学和临床病理学概览，显示蛋白谱与感染的血清学证据密切相关。然而，该研究并非旨在提供以蛋白图谱为中心的深入电泳表型分析。完整电泳图谱随抗利什曼原虫血清学分类升高而变化的程度仍未被充分表征。
本研究旨在在一个接受犬利什曼病（CanL）检测的大型犬队列中，描述抗利什曼原虫血清反应性强度与血清蛋白电泳表型之间的定量关系特征。我们假设，随着血清学分类等级升高，会出现γ-球蛋白进行性扩张、白蛋白浓度及相对比例下降，以及白蛋白与球蛋白比值和白蛋白与γ-球蛋白比值降低。由于电泳图谱还可能受到肾脏蛋白丢失、肝功能障碍和全身性炎症的影响，因此将其解释为综合的临床病理表型，而非仅仅作为体液免疫激活的直接指标。与既往的葡萄牙研究[14]相比，本研究提供了以蛋白图谱为重点的分析，结合了五个有序ELISA分类、序数效应量、探索性受试者工作特征（ROC）分析、校正回归、代表性电泳图以及正式的研究特异性参考区间。由此，本研究提供了一个稳健的定量框架，用于解读与抗利什曼原虫血清学强度升高（而非临床分期疾病严重程度）相关的蛋白图谱改变。在本手稿中，“CanL蛋白图谱”指在CanL诊断和病理生物学背景下所检测的电泳表型，并不意味着每只血清反应阳性犬均患有临床分期疾病。同样，“参考图谱”指定量分布和代表性波形的比较性解读基准，连同源自健康亚群的从头参考区间。次要目标是在一个单独定义的临床健康、抗利什曼原虫血清学阴性的年轻成年犬群体中建立正式的研究特异性参考区间。
2. 材料与方法
2.1. 研究设计与环境
本回顾性、基于实验室的分析性研究使用了2009年至2026年间提交进行利什曼原虫（L. infantum）血清学检测和血清蛋白电泳的家犬记录。样本由位于葡萄牙大陆和自治岛屿地区的兽医医疗中心提交至CEDIVET兽医实验室（葡萄牙波尔图）。
本研究被构思为既往评估CanL流行病学和临床病理学关系工作的以蛋白图谱为中心的延伸。既往研究对在葡萄牙接受检测的犬进行了广泛的流行病学和临床病理学评估，而本研究则专门设计用于评估跨血清学分类等级升高时的血清蛋白电泳表型。该数据集独立于既往发表的队列[14]，无记录层面的重叠，尽管不能排除同一只犬跨实验室就诊的可能性。
在研究期间，共收到来自587家兽医医疗中心（包括诊所和医院）的样本。除生物样本外，转诊兽医还提交了实验室申请单，其中在可获得的情况下包括品种、性别、年龄、采样日期、临床怀疑或检测的流行病学指征以及所要求的实验室分析等信息。
因此，研究群体应被解释为接受实验室检测的犬群体，而非基于群体的患病率样本。犬接受检测的原因包括临床怀疑CanL、居住于或暴露于流行区、接触潜在感染动物、既往或近期前往CanL发生地区旅行，或其他有理由进行利什曼原虫检测的临床或流行病学原因。
血清学状态被用作主要分组变量。本研究按照观察性研究STROBE声明的相关条目进行报告。全文中的“血清阴性”、“血清阳性”和“血清学强度”严格指ELISA分类，并不回顾性地将每只犬归类为感染或临床患病。因此，分析应被解释为抗利什曼原虫血清学强度与血清蛋白电泳表型之间的关联，而非对CanL临床疾病患病率或诊断准确性的估计。
2.2. 纳入标准与研究群体
犬若具有抗利什曼原虫血清学和血清蛋白电泳的可用结果，则符合纳入条件。仅纳入家犬（Canis lupus familiaris）。分析前对记录进行审查以识别重复提交。当检测到同一动物的重复提交时，仅保留按时间顺序第一份具有完整抗利什曼原虫血清学和血清蛋白电泳结果的记录。最终工作数据库包括5544只无重复标识的犬。
血清学结果不确定的犬在使用完整有序血清学量表的分析中作为单独类别保留，但在血清阴性与血清阳性犬之间的二分类比较中被排除。年龄缺失的犬保留在主要电泳分析中，但被排除在需要年龄作为变量的分析之外。
2.3. 样本处理与血清蛋白电泳
所有生物样本在处理前均进行了分析适用性评估。具有可能影响实验室判读的宏观或技术特征的样本，包括明显溶血、脂血、样本量不足、样本类型不适当或其他相关分析前限制因素，均不被视为适合进行相应分析。
血清总蛋白（TP）浓度使用Beckman Coulter AU680e全自动生化分析仪®（Beckman Coulter，Cassina De'Pecchi，意大利）测定。血清蛋白电泳分离采用Sebia® Minicap Flex Piercing分析仪（Sebia，Lisses，法国）按制造商说明书进行毛细管电泳。
电泳图谱包括白蛋白、α-1、α-2、β和γ组分。每个电泳组分均记录相对百分比和绝对浓度。绝对组分浓度由电泳百分比和血清总蛋白浓度推算得出。
分析以下变量：总蛋白、白蛋白百分比、白蛋白浓度、α-1百分比、α-1浓度、α-2百分比、α-2浓度、β百分比、β浓度、γ百分比和γ浓度。还分析了两个衍生比值：白蛋白与球蛋白比值及白蛋白与γ-球蛋白比值。白蛋白与球蛋白比值计算为白蛋白浓度除以计算得出的球蛋白浓度。白蛋白与γ-球蛋白比值计算为白蛋白浓度除以γ-球蛋白浓度。
2.4. 血清学检测与分类
抗利什曼原虫抗体采用基于利什曼原虫（L. infantum）抗原的商用间接ELISA检测，研究期间市售产品为LEISCAN® Leishmania ELISA Test/CIVTEST® Canis Leishmania（HIPRA，Amer，Girona，西班牙），按制造商说明书操作。已发表的LEISCAN®检测独立验证报告灵敏度为92.5%，特异性为100%[34]。
结果以ELISA比值（Rz）值表示，并按制造商截断值系统判读：阴性，Rz ≤ 0.9；不确定，0.9 < Rz ≤ 1.1；低滴度阳性，1.1 < Rz ≤ 1.5；阳性至高滴度阳性，1.5 < Rz ≤ 3；极高滴度阳性，Rz > 3。
二分类分析中，Rz ≤ 0.9的犬归类为血清阴性，Rz > 1.1的犬归类为血清阳性。结果不确定的犬从血清阴性与血清阳性的二分类比较中排除。
2.5. 数据管理与质量控制
所有可用数据从实验室信息系统中提取，并在统计分析前整理于Microsoft Excel®（Microsoft Corporation，Redmond，WA，美国）中。记录从实验室信息系统Sislab®（Glintt，Global Intelligent Technologies，Lisbon，葡萄牙）中提取，其中样本由实验室服务器和软件自动分配唯一识别码。随后对数据库进行重复记录、缺失值、编码不一致及生物学上不合理值的核查。
分析前还审查了蛋白图谱的内部一致性。由于绝对电泳组分浓度由血清总蛋白浓度和电泳组分百分比推算得出，在数据清理过程中评估了总蛋白、相对组分和绝对组分值之间的关系。对极端离群值进行了仔细评估，仅当未发现分析前、分析或数据录入问题且总蛋白、相对组分和绝对组分浓度之间的内部关系保持一致时，才保留相应数值。
2.6. 代表性电泳图谱的选择与制备
选择代表性血清蛋白电泳图谱，以展示抗利什曼原虫血清学分类及 relevant 异常蛋白血症表型中观察到的主要电泳模式。选择基于完整的定量电泳图谱，包括总蛋白浓度、白蛋白、α1-、α2-、β-和γ-球蛋白百分比及浓度，以及白蛋白与球蛋白比值和白蛋白与γ-球蛋白比值。尽管选择时考虑了所有组分，图谱集围绕视觉上最具区分度的β区和γ区表型进行组织。
对于血清学进展图，从分类为阴性、不确定、低滴度阳性、阳性/高滴度阳性和极高滴度阳性的犬中选择代表性电泳图谱。对于图谱集图，额外选择电泳图谱以展示接近正常的图谱、轻度γ-球蛋白扩增、β区扩增、β–γ扩增、严重异常蛋白血症和显著高γ-球蛋白血症。
电泳图谱图像从原始实验室报告中提取，仅进行编辑以去除身份识别信息。未对电泳轨迹本身进行任何修改。代表性图像仅用于视觉展示，未作为独立观察纳入统计分析。由于电泳图谱显示比例可能因报告而异，峰形态的视觉比较应结合相应的定量组分值进行判读。
2.7. 统计分析
数据提取自Sislab®（Glintt, Global Intelligent Technologies，里斯本，葡萄牙），使用Microsoft Excel®（Microsoft Corporation，雷德蒙德，华盛顿州，美国）进行整理，并在分析前检查了重复记录、缺失数据、不合理值和编码不一致。所有统计学处理均使用JMP® 18.0.0版（SAS Institute Inc.，卡里，北卡罗来纳州，美国）和DATAtab®（numiqo e.U.，格拉茨，奥地利，2026）完成。
对人口学、血清学和电泳变量计算了描述性统计量。由于大多数变量呈非正态分布且存在临床偏态，连续变量以中位数、四分位距（IQR）、最小值和最大值进行汇总。分类变量以绝对频率和相对频率进行汇总。
在二分类分析中，排除不确定的血清学结果，比较血清阴性和血清阳性犬。性别与血清学状态之间的关联采用卡方检验评估，效应量以Cramer's V表示。年龄和电泳变量在血清阴性与血清阳性犬之间的比较采用Mann–Whitney U检验，效应量以r报告。
在有序分析中，使用完整的五分类血清学量表：阴性、不确定、低阳性、阳性至高阳性和极高阳性。不同血清学类别间电泳变量的差异采用Kruskal–Wallis检验评估，总体效应量以epsilon平方（ε2）表示。在适当情况下，采用Dunn检验并进行Bonferroni校正进行事后两两比较。血清学类别与电泳变量之间的有序关联采用Spearman秩相关系数（ρ）评估。主要的有序电泳结局为γ-球蛋白浓度，系预先选定，因其最直接反映犬利什曼病（CanL）中的体液免疫激活。
采用ROC曲线分析评估所选电泳变量区分血清阳性与血清阴性犬的单标志物判别性能。报告曲线下面积（AUC）及其95%置信区间（CIs），以评估各电泳变量的判别性能。对于表现最佳的标志物，使用Youden指数（J）确定探索性截断值，并相对于二分类ELISA状态，以灵敏度、特异度以及阳性（LR+）和阴性（LR−）似然比描述操作特征。
为评估ROC分析中判别性能最高的电泳变量与血清学状态之间的关联是否独立于基本人口学和时间协变量，以血清阳性状态为因变量，以所选电泳变量、年龄、性别和采样年份为解释变量，进行二元logistic回归。回归结果以校正比值比（OR）及95% CI报告。作为探索性分析，以五分类血清学量表为因变量，以相同的解释变量进行有序logistic回归，以评估该关联在血清学强度递增的各水平间是否保持一致。报告校正OR及95% CI。
所有检验均为双侧检验，统计学显著性设定为p < 0.05 [35]。然而，结果解读优先考虑效应量、CI、各分析间的一致性及生物学合理性，而非仅依据p值。
2.8. 参考亚群与参考区间估计
在计算区间之前，从接受疫苗接种前抗利什曼原虫检测的独立犬中定义参考亚群。当转诊兽医认定犬临床健康、记录无疾病史或临床证据、ELISA结果为阴性（Rz ≤ 0.9）且采样时年龄≥1岁且<4岁时，该犬符合纳入条件。年龄作为连续变量处理。未推断血液学、生化或尿液分析结果正常。
根据美国兽医临床病理学会（ASVCP）和临床与实验室标准协会（CLSI）的建议[36,37]，中央95%参考区间采用非参数方法从第2.5和第97.5百分位数估计，并计算90% CI。使用Tukey和Dixon–Reed方法筛查潜在离群值，未仅基于统计学理由将其排除。采用非参数分析并结合Benjamini–Hochberg校正评估按性别、年龄和采样时段的潜在分区。
3. 结果
3.1. 研究群体与血清学分类
最终数据集包括5544只接受抗利什曼原虫抗体检测和血清蛋白电泳的犬。其中，2454只犬被分类为血清阴性，180只为不确定，368只为低阳性，1229只为阳性至高阳性，1313只为极高阳性。排除不确定结果后，二分类分析纳入5364只犬，包括2454只血清阴性犬和2910只血清阳性犬。
二元数据集共纳入2074只雌犬和3290只雄犬，所有犬均记录了性别。雌犬中989只血清阴性，1085只血清阳性；雄犬中1465只血清阴性，1825只血清阳性。雌犬血清阳性比例为52.3%，雄犬为55.5%。性别与血清学状态存在统计学关联，但关联强度非常小（χ2 = 5.11，df = 1，p = 0.024；Cramer's V = 0.03）。
在纳入二元血清学分析的5364只犬中，4861只具有年龄数据。其余503只犬（9.4%）的申请文件中未注明年龄，因此被排除在年龄相关分析之外。在具有年龄数据的犬中，年龄范围为0.5岁（6月龄）至22岁。年龄组与血清学状态相关（χ2 = 129.27，df = 4，p < 0.001；Cramer's V = 0.16），血清阳性比例最高的是2至<6岁犬。在连续年龄比较中，血清阳性犬略年轻于血清阴性犬（中位年龄6岁[IQR：4–9] vs. 7岁[IQR：4–10]；Mann–Whitney U = 2,816,360，p = 0.012），但效应量非常小（r = 0.05）。表1显示了按年龄组划分的血清阳性犬比例。
3.2. 血清阴性与血清阳性犬的血清蛋白电泳图谱
与血清阴性犬相比，血清阳性犬的总蛋白、β-球蛋白和γ-球蛋白浓度较高，白蛋白浓度和百分比以及白蛋白/球蛋白比值和白蛋白/γ-球蛋白比值较低（表2）。除α-1-球蛋白浓度（p = 0.483）和β-球蛋白百分比（p = 0.499）外，所有变量在组间差异均显著。年龄差异的p = 0.012，而其他所有显著差异的p < 0.001。效应量最大的是γ-球蛋白浓度（r = 0.462）、γ-球蛋白百分比（r = 0.461）、白蛋白/γ-球蛋白比值（r = 0.453）、白蛋白百分比（r = 0.363）和白蛋白/球蛋白比值（r = 0.363）。
3.3. 按有序血清学分级划分的血清蛋白电泳图谱
对完整的五分类血清学分级进行分析时，γ-球蛋白浓度随分类等级升高而逐渐增加。血清阴性犬的γ-球蛋白浓度中位数为0.8 g/dL，结果不确定犬为0.9 g/dL，低阳性犬为1.0 g/dL，阳性/高阳性犬为1.3 g/dL，极高阳性犬为2.1 g/dL。该有序梯度具有统计学强度和生物学一致性（Kruskal–Wallis χ2 = 1553.91，df = 4，p < 0.001；ε2 = 0.280），表明抗利什曼原虫血清学强度增加伴随γ-球蛋白组分的显著渐进性扩增。
白蛋白/γ-球蛋白比值在相同分类间呈平行下降，阴性、不确定、低阳性、阳性/高阳性和极高阳性犬的中位数分别为4.18（IQR：2.61–6.10）、3.67（2.61–5.25）、3.50（1.99–5.30）、2.36（1.20–3.85）和1.22（0.69–2.21）（Kruskal–Wallis χ2 = 1508.60，df = 4，p < 0.001；ε2 = 0.272）。图2显示了说明该渐进性血清学梯度的代表性电泳图。
Bonferroni校正的Dunn事后比较显示，阴性犬与不确定犬之间差异不显著，不确定犬与低阳性犬在校正后差异也不显著。所有其他两两比较均有统计学显著性。表3总结了按抗利什曼原虫血清学分类划分的γ-球蛋白浓度、95% CI和事后分组。
如表4所示，与抗利什曼原虫血清学分类升高关联最强的有序指标为γ-球蛋白浓度（ρ = 0.522）、γ-球蛋白百分比（ρ = 0.521）、白蛋白/γ-球蛋白比值（ρ = −0.513）、白蛋白百分比（ρ = −0.417）和白蛋白/球蛋白比值（ρ = −0.416）。相比之下，β-球蛋白浓度仅显示弱有序关联（ρ = 0.134），α-1-球蛋白浓度无有意义的有序关联。该模式表明，血清学梯度主要由γ区渐进性扩增和相应的白蛋白相关下降所主导，而非所有球蛋白组分的均匀增加。γ-球蛋白浓度与抗利什曼原虫血清学分类升高的有序关联最强（ρ = 0.522；p < 0.001）。
3.4. 判别性能与校正模型
在单标志物ROC分析中，γ-球蛋白浓度对区分血清阳性与血清阴性犬的判别性能最高，但仅为中等，AUC为0.768（95% CI：0.755–0.780），如图3所示。使用Youden指数，探索性且针对本数据集的γ-球蛋白截断值约为1.05 g/dL，其敏感性为70.2%，特异性为70.5%（LR+ = 2.38；LR− = 0.42）。表5总结了所评估的所有电泳变量的AUC和95% CI。因此，1.05 g/dL值描述的是本数据集中二元ELISA分类的分离情况，并非犬利什曼病的诊断截断值。
总体而言，与抗利什曼原虫血清学阳性关联最密切的电泳变量是γ-球蛋白浓度、γ-球蛋白百分比、白蛋白与γ-球蛋白比值以及白蛋白与球蛋白比值。这一模式支持将蛋白图解释为接受CanL检测犬只体液免疫激活的综合电泳表型。
还选取了代表性电泳图以展示在接受CanL检测犬只中观察到的主要电泳表型（图4）。这些包括接近正常的图谱、轻度γ-球蛋白扩增、β区扩增、β–γ扩增、严重异常蛋白血症和显著高γ-球蛋白血症。这些示例旨在支持蛋白图的视觉解读，应被视为示意性形态学模式，而非独立的统计学观察结果。
在调整年龄、性别和采样年份后，γ-球蛋白浓度与血清学状态之间的关联仍然很强。在二元逻辑回归中，γ-球蛋白浓度每增加1 g/dL，血清阳性的比值增加2.77倍（调整OR = 2.77；95% CI：2.54–3.02；p < 0.001）。年龄较大与血清阳性比值略低相关，而雄性性别和较近的采样年份与比值略高相关。
在使用五分类血清学量表的探索性有序逻辑回归模型中，调整年龄、性别和采样年份后，γ-球蛋白浓度每增加1 g/dL同样与更高的血清学强度相关（调整OR = 2.28；95% CI：2.16–2.41；p < 0.001）。
3.5. 临床健康血清阴性犬的参考区间
参考亚群包括254只独立犬只（149只雄性和105只雌性），均为抗利什曼原虫阴性，年龄≥1至<4岁，且具有完整的电泳数据。年龄分层包括77只年龄≥1至<2岁的犬、95只年龄≥2至<3岁的犬和82只年龄≥3至<4岁的犬。统计分析期间未排除任何观察值。从临床健康血清阴性犬计算的研究特异性95%参考区间见表6，以支持电泳图谱的解读。
26个参考限CI中有22个达到了ASVCP期望的精密度目标。雌性的α-2-球蛋白百分比和浓度略低于雄性（Benjamini–Hochberg校正p < 0.01），而年龄关联较弱（ρ = 0.18）。
4. 讨论
本研究提供了一项以蛋白图为中心的大规模分析，探讨在接受CanL检测的犬只中，血清蛋白毛细管电泳模式如何随抗利什曼原虫抗体滴度升高（以有序ELISA Rz分类反映）而变化。主要发现是在完整的有序血清学量表上，γ-球蛋白区呈现清晰的梯度性扩增，同时伴有白蛋白相关比例以及衍生的白蛋白与球蛋白比值和白蛋白与γ-球蛋白比值的相应降低。这一模式在二元比较、有序分析、效应量估计、ROC分析和调整回归模型中均保持一致，支持将血清蛋白图解释为接受CanL检测犬只体液免疫激活的可测量电泳表型。主要新颖之处在于在此规模的以蛋白图为中心的队列中，结合了完整的五分类血清学量表、效应量估计、调整模型和代表性电泳图。这些发现描述了与血清学强度的关联，不应被解释为临床疾病严重程度增加的证据。
信息量最大的变量是γ-球蛋白浓度。其从血清阴性犬到极高阳性犬的逐步升高表明，电泳图谱在血清阳性动物中不仅仅是异常的，而且随着抗利什曼原虫抗体反应的增强而发生定量变化。这在生物学上是合理的，因为CanL的特征是慢性抗原刺激、体液免疫过度、免疫球蛋白产生和免疫复合物相关病理[16,17,23]。由于IgG主要贡献于γ区，且含IgG的免疫复合物也可能促进γ区扩增，γ-球蛋白浓度与血清学强度升高之间的强关联在生物学上是预期的[12,13,23]。CanL患犬的白蛋白浓度可能通过多种并发机制降低，包括作为负急性期反应一部分的肝脏合成减少、与肾小球疾病或蛋白丢失性肾病相关的肾脏丢失、临床受影响动物营养摄入减少，以及在某些情况下与严重皮肤病变相关的蛋白质丢失[16,17,23,25,26]。白蛋白组分、白蛋白与球蛋白比值和白蛋白与γ-球蛋白比值的平行降低进一步支持了以下观点：与血清学相关的蛋白图反映的是血清蛋白的协调性重新分布，而非单一组分的孤立增加。综上所述，这些发现支持将蛋白图解释为与抗利什曼原虫血清学强度相关的综合临床病理表型，而非仅作为体液免疫的直接测量指标。
这些发现强化了犬利什曼病相关异常蛋白血症的经典描述，包括低白蛋白血症、白蛋白与球蛋白比值降低、多克隆高丙种球蛋白血症以及β区带的不同程度受累[16,17,23,25,29]。然而，本研究在三个重要方面扩展了上述证据。第一，它将蛋白图谱作为主要分析对象而非辅助生化变量进行评估。第二，它采用了五分类血清学量表，使电泳表型能够在抗体负荷的连续谱系上进行解读。第三，大样本量为中位值、CI、序数效应量和代表性电泳表型提供了稳定的定量框架。
正式参考区间组件将图谱扩展到完整实验室检测队列的描述性分布之外。这些区间来源于另行定义的临床健康、血清学阴性的年轻成年犬群体，因此提供了平台特异性基线，可据此解读随血清学强度增加而进展的异常蛋白血症。其精密度总体可接受，26个参考限CI中有22个达到ASVCP理想目标。
γ-球蛋白相关变量的主导地位也具有临床意义。在常规实践中，总蛋白和计算球蛋白值可能提示异常蛋白血症，但无法定位蛋白偏移，也无法区分该变化主要由白蛋白丢失、急性期组分、β区带扩展还是γ区带免疫球蛋白产生所驱动。毛细管电泳提供了这种定位，并将非特异性生化异常转化为更具可解读性的炎症和免疫学特征[3,4,5,6]。在本数据集中，血清学阳性犬的总蛋白升高，但最强的血清学梯度集中在γ区带，表明血清学强度增加的主要电泳特征是体液免疫性的，而非所有组分的普遍性增加。
β组分值得更为谨慎地解读。血清学阳性犬的β-球蛋白浓度较高，但与血清学分类增加仅呈弱序数关联。这提示β区带扩展可能发生于部分犬，可能反映急性期蛋白、补体、脂蛋白、特定免疫球蛋白或β-γ过渡现象，但它并非抗利什曼原虫血清学强度增加的主要标志物。β区带异常为非特异性，也可能受炎症、肝脏或肾脏疾病、脂质代谢、溶血、样本类型和分析方法的影响[3,4,5,6,38]。因此，β扩展应结合形态学和定量进行解读，并始终与完整电泳图谱联合分析。相应地，β区带扩展应被视为次要的非特异性特征，不应赋予与γ区带/白蛋白模式同等的解读权重。
代表性电泳图将定量数据与可识别模式联系起来，但仅具示例性而非诊断性。应结合相应组分值和临床背景进行解读[1,10,13]。
ROC曲线结果应谨慎解读。γ-球蛋白浓度在区分血清学阳性与阴性犬方面取得了最高的单标志物判别性能，但AUC表明为中等而非确定性判别。尽管白蛋白与γ-球蛋白比值综合了白蛋白和γ-球蛋白的相互变化，其AUC略低于单独γ-球蛋白浓度（0.762对0.768）。这可能反映白蛋白的额外变异性，源于与抗体产生无直接关系的肾脏、肝脏、炎症和营养机制。1.05 g/dL值为探索性、数据集特异性阈值，用于区分ELISA分类，不应解读为犬利什曼病的诊断临界值。这一点尤为重要，因为血清蛋白电泳对利什曼原虫感染并非特异性，类似的异常蛋白血症模式也可能见于其他慢性炎症性、感染性、免疫介导性、肝脏、肾脏或肿瘤性疾病[3,4,5,6,32]。因此，蛋白图谱应补充而非替代定量血清学、临床评估、必要时寄生虫检测、尿液分析和肾脏评估[16,17,29]。
结果还需在犬媒介传播疾病的更广泛背景下解读。在对犬埃立克体或嗜吞噬细胞无形体、吉氏巴贝虫或Candidatus Mycoplasma haematoparvum以及D. immitis血清反应阳性的犬中，也出现类似的多克隆或限制性γ区带扩展及白蛋白与球蛋白比值降低[32]。利什曼原虫（L. infantum）感染中也有罕见的单克隆高丙种球蛋白血症伴冷球蛋白血症的报道，可能干扰自动分析仪，进一步强调需要仔细处理样本并在有指征时进行免疫固定电泳[39]。
近期结合血清蛋白电泳与免疫固定电泳对不同媒介传播病原体血清阳性犬的研究显示，多数异常蛋白血症为多克隆性或限制性多克隆性，而真正的M蛋白样模式并不常见[32]。这与犬利什曼病（CanL）直接相关，因为强烈或不规则的免疫球蛋白应答偶尔可产生看似限制性的峰。因此，对于疑似或确诊媒介传播感染的犬，出现窄带或非典型条带时，不应在缺乏适当临床背景的情况下过度解读为肿瘤性副蛋白血症，必要时需进行免疫固定电泳或额外的确证试验。
与既往葡萄牙数据相比，本研究将重点从广泛的临床病理学关联转向电泳表型分析。早先为期15年的回顾性研究显示，γ球蛋白及白蛋白与球蛋白比值是与抗利什曼原虫ELISA结果最为密切相关的实验室变量之一[14]。本分析在更大的以蛋白图谱为重点的队列中证实了该观察结果，并表明该关联具有序数性、定量性和生物学一致性。在此意义上，当前工作应作为既往发现的延伸而非重复来呈现：它将蛋白图谱重新构建为综合的临床病理学读数，并提供了与血清学强度相关的血清蛋白景观的更详细图谱。
4.1. 局限性
本研究为回顾性、基于实验室的研究；因此，样本不能代表一般犬群，也不能用于估计疾病患病率。犬只因临床或流行病学原因被送检，造成选择偏倚，偏向于那些被认为有必要进行利什曼病检测的动物。临床分期、治疗状态、疫苗接种史、定量寄生虫检测、完整的血液学和生化指标、尿液分析、蛋白尿、肾脏分期、急性期蛋白以及合并感染筛查均未能统一获取。这些缺失变量限制了对感染与活动性临床疾病的区分能力，限制了对低白蛋白血症中肾脏或免疫复合物因素的评估能力，也限制了判断个体电泳模式是仅反映利什曼病还是合并疾病的能力。该局限性妨碍了对每只犬进行回顾性临床分类。高定量血清学的公认诊断权重仅适用于临床相容的患者[16,25,30,34]。大样本量提高了该实验室检测队列内的精确度，但并不赋予群体代表性。由于调整模型仅纳入年龄、性别和年份，肾脏或蛋白丢失性疾病、肝脏疾病、全身性炎症、合并感染、营养状态、治疗、疫苗接种、临床分期及其他慢性疾病所致的残余混杂仍可能存在。因此，这些模型估计的是以这三个协变量为条件的关联，并不能确立因果或疾病特异性效应。其余血清阴性犬的详细诊断和炎症状态无法获取，因而无法按感染病因或炎症类别进行可靠分类。
其他分析方面的考量也很重要。毛细管电泳提供高分辨率、自动化的分离，但电泳分数值具有方法特异性，不应不加批判地套用于其他平台或实验室[1,3,4,5,6]。绝对分数浓度由血清总蛋白和相对电泳百分比推算而来，这是标准做法，但使分析前质量和总蛋白测定成为解读的核心。代表性电泳图仅用于教学示意，未作为独立观察进行分析。尽管对年龄、性别和年份的调整支持γ球蛋白关联的稳健性，但在回顾性诊断数据集中仍不能排除残余混杂。
最后，参考亚群还有额外的局限性。临床健康状态依据转诊兽医的疫苗接种前评估和现有病史来定义；同时进行的血液学、生化和尿液分析正常并非强制纳入标准。因此，参考区间仅限于使用同一系统分析的、年龄≥1至<4岁的抗利什曼原虫阴性犬。
4.2. 未来建议
未来的研究应在前瞻性设计中于临床分期的犬中验证这些发现，最好整合定量血清学、寄生虫载量、免疫固定电泳、急性期蛋白、肾脏和肝脏生物标志物、尿液分析、蛋白尿、治疗反应和长期结局。纵向设计将特别有价值，可用于确定治疗后白蛋白、β和γ球蛋白组分变化的速度，特定的电泳轨迹是否能预测复发或持续性肾损伤，以及白蛋白与γ球蛋白比值等联合指标是否能比传统白蛋白与球蛋白比值更好地改善临床监测。此类工作将确定蛋白图谱是否能作为临床分期疾病活动的纵向标志物。前瞻性研究还应比较抗利什曼原虫血清学阴性的犬与确诊病毒、细菌或其他寄生虫感染、非感染性炎症性疾病以及无炎症证据的疾病。此类比较可阐明共有的和疾病相关的电泳模式，并改善对未来诊断数据集的解读。
5. 结论
在这个大型回顾性队列中，抗利什曼原虫抗体滴度升高（反映为较高的ELISA Rz类别）与一致且进展性的血清蛋白电泳表型相关。该表型以γ球蛋白扩增为主，并伴有白蛋白相关比值的相应下降。在评估的电泳变量中，γ球蛋白浓度与ELISA Rz类别的关联最强，对血清阳性提供了中等程度的单标志物区分能力，而β球蛋白浓度显示出较弱的序数关联。这些发现支持血清蛋白毛细管电泳作为一种手段，用于表征在接受CanL检测的犬中与抗利什曼原虫血清学强度相关的综合临床病理表型。定量血清学仍是CanL诊断的核心，高滴度可在相容的临床或临床病理背景下确认疾病。蛋白图谱不应被解读为独立的诊断检测，也不应作为临床严重程度的直接衡量指标。相反，它提供了一个定量和可视化的框架，当与抗体滴度、临床表现、肾脏评估、尿液分析和疾病特异性诊断检测相结合时，可能改善解读。
查看英文原文▾
Serum protein capillary electrophoresis can characterise dysproteinaemia in dogs tested for canine leishmaniosis (CanL) as an integrated clinicopathological phenotype of humoral immune activation. This retrospective laboratory-based study evaluated serum protein electrophoretic patterns in 5544 dogs tested for CanL in Portugal between 2009 and 2026. Anti-Leishmania antibody results were classified into five ordered categories: negative, inconclusive, low positive, positive/high positive and very high positive. Total protein, albumin, alpha-1, alpha-2, beta and gamma fractions, albumin-to-globulin and albumin-to-gamma-globulin ratios were analysed. A separately defined reference subpopulation of 254 clinically healthy seronegative dogs aged ≥ 1 to <4 years was used to establish study-specific non-parametric 95% reference intervals. After excluding inconclusive results, binary analysis included 2454 seronegative and 2910 seropositive dogs. Seropositive dogs had higher median gamma globulins than seronegative dogs (1.5 vs. 0.8 g/dL; p < 0.001), with lower albumin-to-globulin (0.7 vs. 1.0; p < 0.001) and albumin-to-gamma-globulin ratios (1.9 vs. 4.2; p < 0.001). Across the ordinal serological scale, gamma globulins increased from 0.8 to 2.1 g/dL (p < 0.001), whereas the albumin-to-gamma-globulin ratio decreased from 4.2 to 1.2. Gamma globulins showed the strongest ordinal association (ρ = 0.522; p < 0.001) and moderate single-marker discrimination of seropositivity (area under the receiver operating characteristic curve = 0.768; 95% confidence interval: 0.755–0.780). These findings define the CanL proteinogram as a progressive, quantifiable immunopathological phenotype, dominated by gamma-region expansion and reciprocal albumin decline, supporting serum protein electrophoresis as a reference tool for interpretation and monitoring.
1. Introduction
Serum protein electrophoresis is one of the few routine laboratory techniques capable of transforming a simple biochemical abnormality into an interpretable representation of host immunobiology [1,2]. Rather than merely confirming hyperproteinaemia or hyperglobulinaemia, it resolves the serum protein compartment into a structured profile of albumin, alpha (α), beta (β) and gamma (γ) fractions, thereby providing insights into inflammation, acute-phase activation, antigenic stimulation, immunoglobulin production, protein loss and chronic tissue injury [3,4,5,6].
In dogs, serum protein electrophoresis has evolved from a largely descriptive technique for identifying dysproteinaemias into a clinically informative tool for recognising inflammatory, infectious, immune-mediated and neoplastic patterns [4,5,6]. Capillary zone electrophoresis is particularly valuable because of its automation, analytical speed and improved resolution compared with traditional electrophoretic methods [4,7]. However, electrophoretic interpretation remains platform-specific, and results should not be transferred uncritically between analytical systems or laboratories [4,5]. Absolute fraction concentrations are calculated from the electrophoretic percentage and the total serum protein concentration, making the technique both quantitative and morphologically informative [4,5]. Its clinical value, however, depends on interpretation of the complete electrophoretic profile rather than on isolated fraction values.
The biological meaning of each electrophoretic fraction is distinct but interconnected. Albumin reflects hepatic synthesis, oncotic status, negative acute-phase behaviour and protein loss. Alpha fractions include several acute-phase proteins and therefore frequently mirror early or active inflammation [8,9]. Beta fractions comprise a heterogeneous group of proteins, including transferrin, complement, lipoproteins and selected immunoglobulins, and may become particularly relevant in inflammatory, parasitic and immune-complex-associated conditions. In dogs with severe glomerular protein loss, relative prominence of the alpha-2 and beta regions may also occur because high-molecular-weight proteins such as alpha-2-macroglobulin and lipoproteins are comparatively retained. These changes should therefore be interpreted alongside absolute fraction concentrations and renal parameters. The gamma region is predominantly composed of immunoglobulins, particularly immunoglobulin G (IgG), and is therefore closely linked to chronic antigenic stimulation and humoral immune activation [3,4,5,6]. The proteinogram consequently provides a systems-level view of the inflammatory and immune response, integrating biological processes that are often assessed separately through biochemistry, inflammatory markers, serology and clinical staging [8,9,10,11,12,13].
A schematic representation of the normal canine serum proteinogram is provided in Figure 1 to facilitate interpretation of the electrophoretic fractions discussed throughout this study.
Canine leishmaniosis (CanL) is caused predominantly by Leishmania infantum and represents one of the best naturally occurring disease models in veterinary medicine for investigating serum protein electrophoretic alterations associated with chronic inflammation and persistent immune stimulation [14,15]. CanL is a zoonotic vector-borne parasitosis with a wide clinical spectrum, which can range from a subclinical infection to a severe chronic systemic disease [16,17,18,19,20]. Dogs are the main domestic reservoir of L. infantum, and the disease expression depends strongly on the interaction between the parasite and the host immune response [16,21,22]. Protective cellular immunity may contribute to infection control, whereas exaggerated and persistent humoral activation is associated with progressive disease, circulating immune complexes, glomerular injury, systemic inflammation and characteristic dysproteinaemia [16,17,23,24]. The electrophoretic profile in CanL is not a passive laboratory abnormality; it rather constitutes a biochemical expression of the host immunopathological response.
The classical serum protein electrophoretic pattern associated with dogs clinically affected with leishmaniosis includes hypoalbuminaemia, a reduced albumin-to-globulin ratio, polyclonal hypergammaglobulinaemia and, in many dogs, variable beta-globulin expansion and acute-phase changes [16,17,23,25,26]. These findings are biologically consistent with chronic antigenic exposure, non-protective antibody production, immune-complex formation, renal protein loss and inflammatory redistribution of serum proteins [27,28]. None of these abnormalities is pathognomonic for leishmaniosis. Quantitative serology is not merely an exposure marker at all antibody levels: in a dog with compatible clinical or clinicopathological abnormalities, high anti-Leishmania antibody titres may confirm CanL, whereas low titres require further investigation [16,17,29,30]. Accordingly, SPE should not replace quantitative serology or parasite detection; its rationale is to localise dysproteinaemia to the albumin, acute-phase, β- or γ-regions and display broad, restricted or monoclonal-like morphology, thereby adding information unavailable from total protein or calculated globulin alone [3,4,5,6].
The relevance of serum protein electrophoresis in canine vector-borne diseases has recently gained renewed attention. In dogs naturally infected with Dirofilaria immitis, electrophoretic evaluation revealed frequent albumin and beta-globulin abnormalities, as well as treatment-associated improvement in albumin, beta fractions and the albumin-to-globulin ratio, supporting the role of the proteinogram as an adjunctive monitoring tool in parasitic disease [31]. More directly in CanL, a prospective treatment-monitoring study reported that dogs receiving meglumine antimoniate plus allopurinol showed short-term increases in albumin concentration and the albumin-to-globulin ratio, together with a decrease in gamma-globulins, supporting serum protein electrophoresis as a complementary tool for early assessment of therapeutic response [10]. In addition, a recent study evaluating serum protein electrophoresis and immunofixation in dogs seropositive for several vector-borne pathogens showed that most dysproteinaemias were inflammatory or polyclonal in nature, while only a small number of cases fulfilled combined electrophoretic and immunofixation criteria suggestive of a possible M-protein [32]. These observations emphasise two important principles: first, some vector-borne infections can induce characteristic serum protein electrophoretic patterns that reflect the host’s immunopathological response, and second, interpretation of broad, restricted or monoclonal-like bands requires both biological context and analytical caution [6,32,33].
In CanL, however, the proteinogram has often been treated as a descriptive accompaniment to diagnosis rather than as a primary object of investigation. Many studies report hypergammaglobulinaemia, hypoalbuminaemia and a decreased albumin-to-globulin ratio, but relatively few have examined whether the electrophoretic phenotype changes quantitatively across increasing anti-Leishmania serological intensity [14]. This distinction is critical. A binary comparison between seronegative and seropositive dogs may confirm that dysproteinaemia is common in infection, but it does not determine whether the serum protein profile is progressively reshaped as the humoral response intensifies.
A previous large-scale retrospective study in Portugal demonstrated that gamma-globulin concentrations and the albumin-to-globulin ratio were among the clinicopathological variables most clearly associated with anti-Leishmania enzyme-linked immunosorbent assay (ELISA) categories in dogs tested for Leishmania infection [14]. That study provided an important epidemiological and clinicopathological overview, showing that the protein profile was closely linked to serological evidence of infection. However, it was not designed to provide an in-depth, proteinogram-centred analysis of the electrophoretic phenotype. The extent to which the complete electrophoretic profile changes across increasing anti-Leishmania serological categories remains insufficiently characterised.
The present study was designed to characterise the quantitative relationship between the magnitude of anti-Leishmania seroreactivity and the serum protein electrophoretic phenotype in a large cohort of dogs tested for CanL. We hypothesised that increasing serological categories would be associated with progressive gamma-globulin expansion, decreasing albumin concentrations and relative proportions, and declining albumin-to-globulin and albumin-to-gamma-globulin ratios. Because the electrophoretic profile may also be influenced by renal protein loss, hepatic dysfunction and systemic inflammation, it was interpreted as an integrated clinicopathological phenotype rather than solely as a direct measure of humoral immune activation. Compared with the previous Portuguese study [14], the present work provides a proteinogram-focused analysis combining five ordered ELISA categories, ordinal effect sizes, exploratory receiver operating characteristic (ROC) analysis, adjusted regression, representative electropherograms and formal study-specific reference intervals. It thereby provides a robust quantitative framework for interpreting proteinogram alterations associated with increasing anti-Leishmania serological intensity rather than clinically staged disease severity. In this manuscript, “CanL proteinogram” denotes the electrophoretic phenotype examined within the diagnostic and pathobiological context of CanL, it does not imply that every seroreactive dog had clinically staged disease. Likewise, “reference atlas” denotes a comparative interpretive benchmark of quantitative distributions and representative traces, together with de novo reference intervals derived from the healthy subpopulation. A secondary objective was to establish formal study-specific reference intervals in a separately defined population of clinically healthy, anti-Leishmania-seronegative young adult dogs.
2. Materials and Methods
2.1. Study Design and Setting
This retrospective, laboratory-based, analytical study used records from domestic dogs submitted for L. infantum serology and serum protein electrophoresis between 2009 and 2026. Samples were submitted to CEDIVET Veterinary Laboratories (Porto, Portugal) by veterinary medical centres located in mainland Portugal and the autonomous insular regions.
The study was conceived as a proteinogram-centred extension of previous work evaluating epidemiological and clinicopathological relationships in CanL. Whereas the previous study provided a broad epidemiological and clinicopathological assessment in dogs tested in Portugal, the present study was specifically designed to evaluate the serum protein electrophoretic phenotype across increasing serological categories. This dataset is independent of the previously published cohort [14], with no record-level overlap, although cross-laboratory attendance by the same dog cannot be excluded.
During the study period, samples were received from 587 veterinary medical centres, including clinics and hospitals. Along with the biological sample, a laboratory request form was submitted by the referring veterinarian and included, when available, information on breed, sex, age, date of sampling, clinical suspicion or epidemiological indication for testing, and the requested laboratory analyses.
The study population should therefore be interpreted as a laboratory-tested canine population rather than as a population-based prevalence sample. Dogs were tested because of clinical suspicion of CanL, residence in or exposure to endemic areas, contact with potentially infected animals, previous or recent travel to areas where CanL occurs, or other clinical or epidemiological reasons justifying Leishmania testing.
Serological status was used as the main grouping variable. The study was reported in accordance with relevant items of the STROBE statement for observational studies. Throughout, “seronegative”, “seropositive” and “serological intensity” refer strictly to ELISA categories and do not retrospectively classify every dog as infected or clinically diseased. Therefore, the analyses should be interpreted as associations between anti-Leishmania serological intensity and the serum protein electrophoretic phenotype, rather than as estimates of clinical disease prevalence or diagnostic accuracy for CanL.
2.2. Eligibility Criteria and Study Population
Dogs were eligible for inclusion if they had available results for anti-Leishmania serology and serum protein electrophoresis. Only domestic dogs (Canis lupus familiaris) were included. Records were reviewed before analysis to identify repeated submissions. When repeated submissions from the same animal were detected, only the first chronological record with complete anti-Leishmania serology and serum protein electrophoresis results was retained. The final working database included 5544 dogs without repeated identifiers.
Dogs with inconclusive serological results were retained as a separate category in analyses using the complete ordinal serological scale but were excluded from binary comparisons between seronegative and seropositive dogs. Dogs with missing age were retained in the main electrophoretic analyses but excluded from analyses requiring age as a variable.
2.3. Sample Processing and Serum Protein Electrophoresis
All biological samples were assessed for analytical suitability before processing. Samples with macroscopic or technical characteristics likely to compromise laboratory interpretation, including marked haemolysis, lipaemia, inadequate sample volume, inappropriate sample type or other relevant pre-analytical limitations, were not considered suitable for the corresponding analysis.
Serum total protein (TP) concentration was measured using the Beckman Coulter AU680e Automated Chemistry Analyzer® (Beckman Coulter, Cassina De’Pecchi, Italy). Serum protein electrophoretic separation was performed by capillary electrophoresis using the Sebia® Minicap Flex Piercing Analyser (Sebia, Lisses, France), according to the manufacturer’s instructions.
The electrophoretic profile included albumin, alpha-1, alpha-2, beta and gamma fractions. For each electrophoretic fraction, both the relative percentage and the absolute concentration were recorded. Absolute fraction concentrations were derived from the electrophoretic percentage and the total serum protein concentration.
The following variables were analysed: total protein, albumin percentage, albumin concentration, alpha-1 percentage, alpha-1 concentration, alpha-2 percentage, alpha-2 concentration, beta percentage, beta concentration, gamma percentage and gamma concentration. Two derived ratios were also analysed: the albumin-to-globulin ratio and the albumin-to-gamma-globulin ratio. The albumin-to-globulin ratio was calculated as albumin concentration divided by calculated globulin concentration. The albumin-to-gamma-globulin ratio was calculated as albumin concentration divided by gamma-globulin concentration.
2.4. Serological Testing and Classification
Anti-Leishmania antibodies were detected using a commercial indirect ELISA based on L. infantum antigens, commercially available during the study period as LEISCAN® Leishmania ELISA Test/CIVTEST® Canis Leishmania (HIPRA, Amer, Girona, Spain), according to the manufacturer’s instructions. Published independent validation of the LEISCAN® assay reported a sensitivity of 92.5% and a specificity of 100% [34].
Results were expressed as ELISA ratio (Rz) values and interpreted according to the manufacturer’s cut-off system: negative, Rz ≤ 0.9; inconclusive, 0.9 < Rz ≤ 1.1; low positive, 1.1 < Rz ≤ 1.5; positive to high positive, 1.5 < Rz ≤ 3; and very high positive, Rz > 3.
For binary analyses, dogs were classified as seronegative when Rz ≤ 0.9 and seropositive when Rz > 1.1. Dogs with inconclusive results were excluded from binary seronegative versus seropositive comparisons.
2.5. Data Management and Quality Control
All available data were extracted from the laboratory information system and organised in Microsoft Excel® (Microsoft Corporation, Redmond, WA, USA) before statistical analysis. Records were extracted from the laboratory information system, Sislab® (Glintt, Global Intelligent Technologies, Lisbon, Portugal), in which samples are automatically assigned unique identification codes by the laboratory server and software. The database was then checked for duplicate records, missing values, coding inconsistencies, and biologically implausible values.
The internal coherence of the proteinogram was also reviewed before analysis. Because absolute electrophoretic fraction concentrations were derived from total serum protein concentration and electrophoretic fraction percentage, the relationships between total protein, relative fractions and absolute fraction values were evaluated during data cleaning. The presence of extreme outliers was carefully evaluated, and values were retained only when no pre-analytical, analytical or data-entry issue was identified and the internal relationships between total protein, relative fractions and absolute fraction concentrations remained coherent.
2.6. Selection and Preparation of Representative Electropherograms
Representative serum protein electropherograms were selected to illustrate the main electrophoretic patterns observed across anti-Leishmania serological categories and relevant dysproteinaemic phenotypes. Selection was based on the complete quantitative electrophoretic profile, including total protein concentration, albumin, α1-, α2-, β- and γ-globulin percentages and concentrations, and the albumin-to-globulin and albumin-to-γ-globulin ratios. Although all fractions were considered during selection, the atlas was organised around the most visually distinctive β- and γ-region phenotypes.
For the serological progression figure, representative electropherograms were selected from dogs classified as negative, inconclusive, low positive, positive/high positive and very high positive. For the atlas figure, additional electropherograms were selected to illustrate near-normal profiles, mild γ-globulin expansion, β-region expansion, β–γ expansion, severe dysproteinaemia and marked hypergammaglobulinaemia.
Electropherogram images were extracted from original laboratory reports and edited only to remove identifying information. No modification was made to the electrophoretic trace itself. The representative images were used for visual illustration only and were not included as independent observations in the statistical analyses. Because electropherogram display scaling may vary between reports, visual comparison of peak morphology should be interpreted together with the corresponding quantitative fraction values.
2.7. Statistical Analysis
Data were extracted from Sislab® (Glintt, Global Intelligent Technologies, Lisbon, Portugal), organised in Microsoft Excel® (Microsoft Corporation, Redmond, WA, USA) and checked for duplicate records, missing data, implausible values and coding inconsistencies before analysis. All statistical procedures were performed using JMP® version 18.0.0 (SAS Institute Inc., Cary, NC, USA) and DATAtab® (numiqo e.U., Graz, Austria, 2026).
Descriptive statistics were calculated for demographic, serological and electrophoretic variables. Continuous variables were summarised as median, interquartile range (IQR), minimum and maximum because most variables were non-normally distributed and clinically skewed. Categorical variables were summarised as absolute and relative frequencies.
For binary analyses, inconclusive serological results were excluded, and seronegative and seropositive dogs were compared. The association between sex and serological status was assessed using the chi-square test, with effect size expressed as Cramer’s V. Age and electrophoretic variables were compared between seronegative and seropositive dogs using the Mann–Whitney U test, with effect size reported as r.
For ordinal analyses, the complete five-category serological scale was used: negative, inconclusive, low positive, positive to high positive and very high positive. Differences in electrophoretic variables across serological categories were assessed using the Kruskal–Wallis test, with omnibus effect size expressed as epsilon-squared (ε2). When appropriate, post hoc pairwise comparisons were performed using Dunn’s tests with Bonferroni adjustment. Ordinal associations between serological category and electrophoretic variables were assessed using Spearman’s rank correlation coefficient (ρ). The primary ordinal electrophoretic outcome was γ-globulin concentration, selected a priori because it most directly reflects humoral immune activation in CanL.
ROC curve analysis was used to evaluate the single-marker discriminatory performance of selected electrophoretic variables for differentiating seropositive from seronegative dogs. The area under the curve (AUC) was reported with 95% confidence intervals (CIs) to assess discriminatory performance of each electrophoretic variable. For the best-performing marker, exploratory cut-off values were identified using the Youden index (J), and operating characteristics relative to binary ELISA status were described using sensitivity, specificity, and positive (LR+) and negative (LR−) likelihood ratios.
To assess whether the association between the electrophoretic variable showing the highest discriminatory performance in ROC analysis and serological status was independent of basic demographic and temporal covariates, binary logistic regression was performed with seropositive status as the dependent variable and the selected electrophoretic variable, age, sex and year of sampling as explanatory variables. Regression results were reported as adjusted odds ratios (OR) with 95% CI. As an exploratory analysis, ordinal logistic regression was performed using the five-category serological scale as the dependent variable and the same explanatory variables to evaluate whether the association remained consistent across increasing levels of serological intensity. Adjusted OR with 95% CI were reported.
All tests were two-sided, and statistical significance was set at p < 0.05 [35]. However, interpretation prioritised effect size, CI, consistency across analyses and biological plausibility rather than p-values alone.
2.8. Reference Subpopulation and Reference-Interval Estimation
A reference subpopulation was defined before interval calculation from unique dogs submitted for pre-vaccination anti-Leishmania testing. Dogs were eligible when the referring veterinarian identified them as clinically healthy, recorded no history or clinical evidence of disease, the ELISA result was negative (Rz ≤ 0.9) and age at sampling was ≥1 and <4 years. Age was treated as a continuous variable. Normal haematological, biochemical or urinalysis results were not inferred.
In accordance with American Society for Veterinary Clinical Pathology (ASVCP) and Clinical and Laboratory Standards Institute (CLSI) recommendations [36,37], central 95% reference intervals were estimated non-parametrically from the 2.5th and 97.5th percentiles, with 90% CI. Potential outliers were screened using Tukey and Dixon–Reed methods and were not excluded solely on statistical grounds. Potential partitioning by sex, age and sampling period was assessed using non-parametric analyses with Benjamini–Hochberg adjustment.
3. Results
3.1. Study Population and Serological Classification
The final dataset included 5544 dogs tested for anti-Leishmania antibodies and serum protein electrophoresis. Of these, 2454 dogs were classified as seronegative, 180 as inconclusive, 368 as low positive, 1229 as positive to high positive and 1313 as very high positive. After exclusion of inconclusive results, the binary analysis included 5364 dogs, comprising 2454 seronegative dogs and 2910 seropositive dogs.
Sex was recorded for all dogs included in the binary dataset, which comprised 2074 females and 3290 males. Among females, 989 were seronegative and 1085 were seropositive. Among males, 1465 were seronegative and 1825 were seropositive. The proportion of seropositive dogs was 52.3% in females and 55.5% in males. Sex was statistically associated with serological status, although the magnitude of association was very small (χ2 = 5.11, df = 1, p = 0.024; Cramer’s V = 0.03).
From the 5364 dogs included in the binary serological analysis, age data were available for 4861 individuals. For the remaining 503 dogs (9.4%), age was not specified in the requisition file, and these were therefore excluded from age-related analyses. Among dogs with available age data, age ranged from 0.5 (6 months) to 22 years. Age group was associated with serological status (χ2 = 129.27, df = 4, p < 0.001; Cramer’s V = 0.16), with the highest seropositive proportion observed in dogs aged 2 to <6 years. In the continuous-age comparison, seropositive dogs were slightly younger than seronegative dogs (median age 6 years [IQR: 4–9] vs. 7 years [IQR: 4–10]; Mann–Whitney U = 2,816,360, p = 0.012), although the effect size was very small (r = 0.05). Table 1 displays the proportion of seropositive dogs according to age group.
3.2. Serum Protein Electrophoretic Profile in Seronegative and Seropositive Dogs
Compared with seronegative dogs, seropositive dogs had higher total protein, beta-globulin and gamma-globulin concentrations, lower albumin concentration and percentage, and lower albumin-to-globulin and albumin-to-gamma-globulin ratios (Table 2). All variables differed significantly between groups, except alpha-1-globulin concentration (p = 0.483) and beta-globulin percentage (p = 0.499). The difference in age had p = 0.012, whereas all other significant differences had p < 0.001. The largest effect sizes were observed for gamma-globulin concentration (r = 0.462), gamma-globulin percentage (r = 0.461), the albumin-to-gamma-globulin ratio (r = 0.453), albumin percentage (r = 0.363) and the albumin-to-globulin ratio (r = 0.363).
3.3. Serum Protein Electrophoretic Profile Across the Ordinal Serological Scale
When the complete five-category serological scale was analysed, gamma-globulin concentration increased progressively across categories. Median gamma-globulin concentration was 0.8 g/dL in seronegative dogs, 0.9 g/dL in inconclusive dogs, 1.0 g/dL in low-positive dogs, 1.3 g/dL in positive/high-positive dogs and 2.1 g/dL in very high-positive dogs. This ordinal gradient was statistically strong and biologically coherent (Kruskal–Wallis χ2 = 1553.91, df = 4, p < 0.001; ε2 = 0.280), indicating that increasing anti-Leishmania serological intensity was accompanied by a marked progressive expansion of the gamma-globulin fraction.
The albumin-to-gamma-globulin ratio declined in parallel across the same categories, with medians of 4.18 (IQR: 2.61–6.10), 3.67 (2.61–5.25), 3.50 (1.99–5.30), 2.36 (1.20–3.85) and 1.22 (0.69–2.21) in negative, inconclusive, low-positive, positive/high-positive and very high-positive dogs, respectively (Kruskal–Wallis χ2 = 1508.60, df = 4, p < 0.001; ε2 = 0.272). Representative electropherograms illustrating this progressive serological gradient are shown in Figure 2.
Bonferroni-adjusted Dunn’s post hoc comparisons showed that negative and inconclusive dogs did not differ significantly, and that inconclusive and low-positive dogs also did not differ significantly after adjustment. All other pairwise comparisons were statistically significant. Table 3 summarises gamma-globulin concentration, 95% CI and post hoc groupings according to anti-Leishmania serological category.
As shown in Table 4, the strongest ordinal associations with increasing anti-Leishmania serological category were observed for gamma-globulin concentration (ρ = 0.522), γ-globulin percentage (ρ = 0.521), albumin-to-gamma-globulin ratio (ρ = −0.513), albumin percentage (ρ = −0.417) and albumin-to-globulin ratio (ρ = −0.416). In contrast, beta-globulin concentration showed only a weak ordinal association (ρ = 0.134), and alpha-1-globulin concentration showed no meaningful ordinal association. This pattern indicates that the serological gradient was dominated by progressive γ-region expansion and reciprocal albumin-related decline, rather than by a uniform increase in all globulin fractions. Gamma-globulin concentration showed the strongest ordinal association with increasing anti-Leishmania serological category (ρ = 0.522; p < 0.001).
3.4. Discriminatory Performance and Adjusted Modelling
In single-marker ROC analysis, gamma-globulin concentration showed the highest, but moderate, discriminatory performance for differentiating seropositive from seronegative dogs, with an AUC of 0.768 (95% CI: 0.755–0.780) as shown in Figure 3. Using the Youden index, an exploratory and dataset-specific gamma-globulin cut-off of approximately 1.05 g/dL yielded a sensitivity of 70.2% and a specificity of 70.5% (LR+ = 2.38; LR− = 0.42). Table 5 summarises the AUC and 95% CI for all electrophoretic variables evaluated. The 1.05 g/dL value therefore describes separation of binary ELISA categories within this dataset and is not a diagnostic cut-off for CanL.
Overall, the electrophoretic variables most closely associated with anti-Leishmania serological positivity were gamma-globulin concentration, gamma-globulin percentage, the albumin-to-gamma-globulin ratio and the albumin-to-globulin ratio. This pattern supports the interpretation of the proteinogram as an integrated electrophoretic phenotype of humoral immune activation in dogs tested for CanL.
Representative electropherograms were also selected to illustrate the main electrophoretic phenotypes observed in dogs tested for CanL (Figure 4). These included near-normal profiles, mild gamma-globulin expansion, β-region expansion, β–γ expansion, severe dysproteinaemia and marked hypergammaglobulinaemia. These examples are intended to support visual interpretation of the proteinogram and should be regarded as illustrative morphologic patterns rather than independent statistical observations.
The association between gamma-globulin concentration and serological status remained strong after adjustment for age, sex and year of sampling. In binary logistic regression, each 1 g/dL increase in gamma-globulin concentration was associated with a 2.77-fold increase in the odds of seropositivity (adjusted OR = 2.77; 95% CI: 2.54–3.02; p < 0.001). Older age was associated with slightly lower odds of seropositivity, whereas male sex and more recent years of sampling were associated with slightly higher odds.
In an exploratory ordinal logistic regression model using the five-category serological scale, each 1 g/dL increase in gamma-globulin concentration was likewise associated with higher serological intensity after adjustment for age, sex, and year of sampling (adjusted OR = 2.28; 95% CI: 2.16–2.41; p < 0.001).
3.5. Reference Intervals in Clinically Healthy Seronegative Dogs
The reference subpopulation comprised 254 unique dogs (149 males and 105 females), all anti-Leishmania negative, aged ≥ 1 to <4 years and with complete electrophoretic data. The age strata contained 77 dogs aged ≥ 1 to <2 years, 95 aged ≥ 2 to <3 years and 82 aged ≥ 3 to <4 years. No observations were excluded during statistical analysis. The study-specific 95% reference intervals calculated from clinically healthy seronegative dogs are presented in Table 6 to support interpretation of the electrophoretic profiles.
The desirable ASVCP precision target was met for 22 of 26 reference-limit CIs. Females had modestly lower alpha-2-globulin percentage and concentration than males (Benjamini–Hochberg-adjusted p < 0.01), while age associations were weak (ρ = 0.18).
4. Discussion
This study provides a large-scale, proteinogram-centred analysis of how serum protein capillary electrophoretic patterns vary across increasing anti-Leishmania antibody titres, as reflected by the ordinal ELISA Rz categories, in dogs tested for CanL. The principal finding was a clear, graded expansion of the gamma-globulin region across the complete ordinal serological scale, accompanied by reciprocal reductions in albumin-related proportions and derived albumin-to-globulin and albumin-to-gamma-globulin ratios. This pattern was consistent in binary comparisons, ordinal analyses, effect-size estimates, ROC analysis and adjusted regression modelling, supporting the interpretation that the serum proteinogram captures a measurable electrophoretic phenotype of humoral immune activation in dogs tested for CanL. The principal novelty lies in combining the complete five-category serological scale, effect-size estimates, adjusted models and representative electropherograms in a proteinogram-centred cohort of this size. These findings describe an association with serological intensity and should not be interpreted as evidence of increasing clinical disease severity.
The most informative variable was gamma-globulin concentration. Its progressive increase from seronegative to very high-positive dogs indicates that the electrophoretic profile is not merely abnormal in seropositive animals, but changes quantitatively as the anti-Leishmania antibody response intensifies. This is biologically plausible, because CanL is characterised by chronic antigenic stimulation, exaggerated humoral immunity, immunoglobulin production and immune-complex-associated pathology [16,17,23]. Because IgG contributes predominantly to the γ-region and IgG-containing immune complexes may also contribute to γ-region expansion, the strong association between gamma-globulin concentration and increasing serological intensity is biologically expected [12,13,23]. Albumin concentration may decrease in dogs with CanL through several concurrent mechanisms, including reduced hepatic synthesis as part of the negative acute-phase response, renal loss associated with glomerular disease or protein-losing nephropathy, reduced nutritional intake in clinically affected animals and, in some cases, protein loss associated with severe cutaneous lesions [16,17,23,25,26]. The parallel decrease in albumin fraction, albumin-to-globulin ratio and albumin-to-gamma-globulin ratio further supports the view that the serology-associated proteinogram reflects a coordinated redistribution of serum proteins rather than an isolated increase in one fraction. Taken together, these findings support interpretation of the proteinogram as an integrated clinicopathological phenotype associated with anti-Leishmania serological intensity, rather than as a direct measure of humoral immunity alone.
These findings reinforce the classical description of CanL-associated dysproteinaemia, which includes hypoalbuminaemia, reduced albumin-to-globulin ratio, polyclonal hypergammaglobulinaemia and variable β-region involvement [16,17,23,25,29]. However, the present study extends that evidence in three important ways. First, it evaluates the proteinogram as the primary object of analysis rather than as an ancillary biochemical variable. Second, it uses a five-category serological scale, allowing the electrophoretic phenotype to be interpreted along a continuum of antibody burden. Third, the large sample size provides a stable quantitative framework for median values, CI, ordinal effect sizes and representative electrophoretic phenotypes.
The formal reference-interval component extends the atlas beyond descriptive distributions in the complete laboratory-tested cohort. The intervals were derived from a separately defined population of clinically healthy, seronegative young adult dogs and therefore provide a platform-specific baseline against which the progressive dysproteinaemia associated with increasing serological intensity can be interpreted. Their precision was generally acceptable, with 22 of 26 reference-limit CIs meeting the desirable ASVCP target.
The dominance of gamma-globulin-related variables is also clinically meaningful. In routine practice, total protein and calculated globulin values may suggest dysproteinaemia, but they cannot localise the protein shift or distinguish whether the change is driven mainly by albumin loss, acute-phase fractions, β-region expansion or γ-region immunoglobulin production. Capillary electrophoresis provides that localisation and transforms a non-specific biochemical abnormality into a more interpretable inflammatory and immunological profile [3,4,5,6]. In the present dataset, total protein was increased in seropositive dogs, but the strongest serological gradient was concentrated in the γ-region, indicating that the main electrophoretic signature of increasing serological intensity was humoral rather than a generalised increase in all fractions.
The beta fraction deserves a more cautious interpretation. Beta-globulin concentration was higher in seropositive dogs but showed only a weak ordinal association with increasing serological category. This suggests that β-region expansion may occur in a subset of dogs, possibly reflecting acute-phase proteins, complement, lipoproteins, selected immunoglobulins or beta-gamma transition phenomena, but it was not the dominant marker of increasing anti-Leishmania serological intensity. Beta-region abnormalities are non-specific and may also be influenced by inflammation, hepatic or renal disease, lipid metabolism, haemolysis, sample type and analytical method [3,4,5,6,38]. Therefore, beta expansion should be interpreted morphologically and quantitatively, and always in conjunction with the complete electrophoretic profile. Accordingly, β-region expansion should be regarded as a secondary, non-specific feature and should not be assigned interpretive weight equivalent to the γ-region/albumin pattern.
The representative electropherograms link quantitative data to recognisable patterns but are illustrative rather than diagnostic. They should be interpreted together with the corresponding fraction values and clinical context [1,10,13].
ROC curve results should be interpreted with restraint. Gamma-globulin concentration achieved the highest single-marker discriminatory performance for differentiating seropositive from seronegative dogs, but the AUC indicates moderate rather than definitive discrimination. Although the albumin-to-gamma-globulin ratio combines reciprocal changes in albumin and gamma-globulins, its AUC was slightly lower than that of γ-globulin concentration alone (0.762 versus 0.768). This may reflect additional variability in albumin arising from renal, hepatic, inflammatory and nutritional mechanisms unrelated directly to antibody production. The 1.05 g/dL value is an exploratory, dataset-specific threshold for separating ELISA categories and should not be interpreted as a diagnostic cut-off for CanL. This is especially important because serum protein electrophoresis is not specific for Leishmania infection, and similar dysproteinaemic patterns may occur in other chronic inflammatory, infectious, immune-mediated, hepatic, renal or neoplastic conditions [3,4,5,6,32]. Accordingly, the proteinogram should complement, not replace, quantitative serology, clinical assessment, parasite detection when indicated, urinalysis and renal evaluation [16,17,29].
The results also need to be interpreted within the broader context of canine vector-borne diseases. Analogous polyclonal or restricted γ-region expansion and reduced albumin-to-globulin ratios occur in dogs seroreactive to Ehrlichia canis or Anaplasma phagocytophilum, Babesia gibsoni or Candidatus Mycoplasma haematoparvum, and D. immitis [32]. Rare monoclonal hypergammaglobulinaemia with cryoglobulinaemia has also been reported in L. infantum infection and may interfere with automated analysers, reinforcing the need for careful sample handling and, where indicated, immunofixation [39].
Recent work combining serum protein electrophoresis and immunofixation in dogs seropositive for different vector-borne pathogens showed that most dysproteinaemias were polyclonal or restricted polyclonal in nature, whereas true M-protein-like patterns were uncommon [32]. This is directly relevant to CanL because intense or irregular immunoglobulin responses may occasionally produce restricted-looking peaks. Therefore, narrow or atypical bands in dogs with suspected or confirmed vector-borne infection should not be overinterpreted as neoplastic paraproteinaemia without appropriate clinical context and, when necessary, immunofixation or additional confirmatory testing.
Compared with previous Portuguese data, this study shifts the emphasis from broad clinicopathological association to electrophoretic phenotyping. The earlier 15-year retrospective study showed that gamma globulins and albumin-to-globulin ratio were among the laboratory variables most closely associated with anti-Leishmania ELISA results [14]. The present analysis confirms that observation in a larger proteinogram-focused cohort and demonstrates that the association is ordinal, quantitative and biologically coherent. In this sense, the current work should be presented as an extension rather than a repetition of previous findings: it reframes the proteinogram as an integrated clinicopathological readout and provides a more detailed atlas of the serum protein landscape associated with serological intensity.
4.1. Limitations
The study was retrospective and laboratory-based; therefore, the sample does not represent the general canine population and cannot be used to estimate disease prevalence. Dogs were submitted for clinical or epidemiological reasons, creating selection bias towards animals in which leishmaniosis testing was considered relevant. Clinical staging, treatment status, vaccination history, quantitative parasite detection, complete haematological and biochemical profiles, urinalysis, proteinuria, renal staging, acute-phase proteins and screening for co-infections were not uniformly available. These missing variables limit the ability to distinguish infection from active clinical disease, to assess renal or immune-complex contributions to hypoalbuminaemia, or to determine whether individual electrophoretic patterns reflect leishmaniosis alone or concurrent disease. This limitation prevents retrospective clinical classification of each dog. The recognised diagnostic weight of high quantitative serology applies only in a clinically compatible patient [16,25,30,34]. The large sample size improves precision within this laboratory-tested cohort but does not confer population representativeness. Residual confounding by renal or protein-losing disease, hepatic disease, systemic inflammation, co-infections, nutritional status, treatment, vaccination, clinical stage and other chronic diseases remains possible because only age, sex and year were included in the adjusted models. Accordingly, the models estimate associations conditional on those three covariates and do not establish causal or disease-specific effects. Detailed diagnoses and inflammatory status were unavailable for the remaining seronegative dogs, precluding reliable classification by infectious aetiology or inflammatory category.
Additional analytical considerations are also important. Capillary electrophoresis provides high-resolution, automated separation, but electrophoretic fraction values are method-specific and should not be transferred uncritically to other platforms or laboratories [1,3,4,5,6]. Absolute fraction concentrations were derived from total serum protein and relative electrophoretic percentages, which is standard practice but makes pre-analytical quality and total protein measurement central to interpretation. The representative electropherograms were selected for didactic illustration and were not analysed as independent observations. Although adjustment for age, sex and year supported the robustness of the gamma-globulin association, residual confounding cannot be excluded in a retrospective diagnostic dataset.
Finally, the reference subpopulation has additional limitations. Clinical health was defined from the referring veterinarian’s pre-vaccination assessment and available history; normal concurrent haematology, biochemistry and urinalysis were not mandatory inclusion criteria. The intervals are consequently restricted to anti-Leishmania-negative dogs aged ≥ 1 to <4 years analysed using the same system.
4.2. Future Recommendations
Future studies should validate these findings prospectively in clinically staged dogs, ideally integrating quantitative serology, parasite load, immunofixation, acute-phase proteins, renal and hepatic biomarkers, urinalysis, proteinuria, treatment response and long-term outcomes. Longitudinal designs would be particularly valuable to determine how rapidly albumin, beta and gamma fractions change after treatment, whether specific electrophoretic trajectories predict relapse or persistent renal damage, and whether combined indices such as the albumin-to-gamma-globulin ratio improve clinical monitoring beyond the conventional albumin-to-globulin ratio. Such work would determine whether the proteinogram can serve as a longitudinal marker of clinically staged disease activity. Prospective studies should also compare anti-Leishmania-seronegative dogs with confirmed viral, bacterial or other parasitic infections, non-infectious inflammatory conditions, and conditions without evidence of inflammation. Such comparisons could clarify shared and disease-associated electrophoretic patterns and improve interpretation of future diagnostic datasets.
5. Conclusions
In this large retrospective cohort, increasing anti-Leishmania antibody titres, as reflected by higher ELISA Rz categories, were associated with a coherent and progressive serum protein electrophoretic phenotype. This phenotype was dominated by gamma-globulin expansion, accompanied by a reciprocal decline in albumin-related ratios. Among the electrophoretic variables evaluated, gamma-globulin concentration showed the strongest association with ELISA Rz category and provided moderate single-marker discrimination of seropositivity, whereas beta-globulin concentration showed a weaker ordinal association. These findings support serum protein capillary electrophoresis as a means of characterising an integrated clinicopathological phenotype associated with anti-Leishmania serological intensity in dogs tested for CanL. Quantitative serology remains central to CanL diagnosis, and high titres may confirm disease in a compatible clinical or clinicopathological setting. The proteinogram should not be interpreted as a stand-alone diagnostic test, nor as a direct measure of clinical severity. Rather, it provides a quantitative and visual framework that may refine interpretation when integrated with antibody titre, clinical findings, renal assessment, urinalysis and disease-specific diagnostic testing.