Protoporphyrin (PPIX) and its precursor 5-aminolevulinic acid (ALA) are routinely used in fluorescence-guided resection (FGR) in neurosurgery of glioblastoma. Orally administered prior to surgery, ALA diffuses preferentially into the tumour-bearing brain region and is there transformed into PPIX as part of the heme biosynthesis. During the resection of high-grade gliomas (HGG), the fluorescent properties of PPIX enable the differentiation of healthy and malignant brain tissue – resulting in a more complete tumour resection and, ultimately, a better prognosis for patients.^{1, 2}

Despite the use of ALA-FGR, complete tumour removal is impossible – due to the infiltrative growth of HGG, e.g., glioblastoma multiforme (GBM), tumours often reoccur. Clinical diagnosis of HGG/GBM is furthermore not straightforward and requires expensive imaging technology as well as tissue biopsies. So far, the prognosis for GBM patients remains one of the lowest in modern day oncology.^{1, 3, 4}

We thus investigated PPIX as a potential blood biomarker for the diagnosis of primary and recurrent GBM using liquid chromatography/mass spectrometry (LC/MS).⁵ Some cancer research studies to that effect using plasma have already been published.^6-8 It was also demonstrated that extracellular vesicles originating from glioma cells contained PPIX,⁹ but none of these investigations used mass spectrometry (MS) as detector.

Whole blood contains the most PPIX, but it is not the best matrix for such studies, because of unspecific contributions from erythrocyte ZnPPIX. For serum or plasma, though, the published PPIX levels, obtained with traditional methods, are low, often below the limit of detection, and not reliable.^{10, 11} Taking advantage of the high sensitivity of MS, we have been aiming for a reproducible LC/MS workflow for the use of serum and tested our protocol in comparison to whole blood.⁵ Thereby, the reliable purification, respective recovery, of PPIX from the biological matrix proved particularly difficult and we have therefore evaluated different preparation methods. Initial liquid–liquid extraction (LLE) with water and acetonitrile (ACN) was crucial for best sensitivity.^{5, 12} However, an additional purification step was required to minimize LC-column aging and clogging.

We first tried to separate the small molecular weight substances from macromolecules using syringe and centrifugal filters. It soon turned out, however, that of some of the membrane material we tested – polytetrafluoroethylene, polyvinylidene fluoride, modified polyether sulfone, nylon and cellulose – in particular nylon (polyamide 6.6) and cellulose – quantitatively adsorbed PPIX. This observation led to experiments using punched-out membrane pieces for the specific extraction of PPIX directly from the biological matrix, because PPIX could subsequently be easily removed from the membrane using dimethyl sulfoxide (DMSO). The results from crude preliminary experiments were indeed promising considering a recovery of about 60–70% for PPIX from a reference extract, but handling was difficult, and reproducibility was hampered by the need to produce and process small to tiny pieces of nylon. The ratio between the amount of porphyrin and the nylon surface area was crucial as well as the pH during sample loading and the elution volume. Moreover, the extraction of mesoporphyrin (MPIX), our internal standard, from spiked serum was not as efficient as that of PPIX when working with serum instead of standard solution (Figure 1). At that point we switched to anionic-exchange solid-phase extraction (ae-SPE), but we did feel that the potential of nylon with respect to PPIX isolation should be kept in mind.

Ultimately, ae-SPE worked best for our purpose, but, interestingly, not all sorbent material performed with the same quality. We tested three products with different types of ae-sorbent. All cartridge bed materials consisted of quaternary amine functionalized polymers with strong ae-properties; they differed only in the polymer and the particle size (Table 1). We evaluated a reference solution mimicking the LLE extract and spiked whole blood as well as native and spiked serum. The recovery on styrene (Sty)-based cartridges was the lowest for all matrices (Figure 2) when using a typical SPE workflow including bed preparation, ion exchange, analyte washing and elution (Figure 3) and thus the determination of native PPIX in serum was not attempted with this material. The other two cartridge types performed similarly with recoveries above 80% for spiked PPIX and MPIX (Figures 2A and 2B).

Fully water-wettable polymeric-based cartridges possess the advantage, compared with silica and partially water-wettable material, that an aqueous sample can be loaded directly without performing sorbent conditioning and equilibration steps (Figure 3)¹³ and for the hydrophilic–lipophilic balanced (HLB) copolymer this has been suggested by the manufacturer. Furthermore, polymer-based cartridges can be allowed to run dry – as easily happens during vacuum-based handling – without any negative impact on performance. This simplifies ae-SPE and saves solvent and time. Comparing this shorter workflow for all cartridge types (Figures 2C and 2D), again the HLB copolymer and the mixed-mode (MM) ae-sorbent outperformed the Sty-based cartridge type, with the former fairing slightly better than the latter. No loss in recovery or reproducibility and no differences in extract purity were noticed for all of them. The determination of native PPIX from serum was possible with all three cartridge types. The averaged relative standard deviation (RSD) for triplicate LC/MS runs of MPIX and PPIX was 8 and 5%, respectively. The averaged RSD of ae-SPE (n = 2) of MPIX ranged from 3% on HLB to 6% on MMae cartridges and for PPIX it was 6% for all products. In an extension to the work of Zhang and co-workers,¹³ our results demonstrated that not only the HLB-based sorbent was amenable to the shortened SPE protocol.

Another decisive factor was the ease of use when working with a vacuum station. It was not recommended to exceed a flow rate of about one drop per second¹⁴ in order to allow sufficient interaction between sorbent and analyte, because, in comparison with normal- or reversed-phase interactions, the kinetic exchange process in the ion-exchange process is slower. While HLB and Sty-based cartridges were easily run with the recommended flow rate, adjusting the flow for MMae cartridges was laborious and time-consuming for no obvious reason.

Importantly, ZnPPIX and PPIX could be properly distinguished during ae-SPE, because MPIX and PPIX eluted at 2% formic acid (FA) and ZnPPIX did not up to a concentration of 20% FA.⁵ Considering all parameters, we have thus voted for the use of the HLB material for our study. For the extraction of low-abundance endogenous PPIX from serum (500 μL), containing about 2–3 pmol PPIX/mL in healthy volunteers, and whole blood this sorbent allowed the shortest possible workflow as well as easy handling and it outperformed – even if only marginally in one case – the competing products. We will extend the use of our procedure to tumour tissue. Although not demonstrated here, we expect it to perform equally well when purifying PPIX from LLE extracts of tissue homogenates.

Experimental: Adsorption experiments were performed using round disks made of nylon membrane (Roth, Karlsruhe, Germany) with a surface area of 20 mm² that were manually punched-out with a metal hollow punch. The disks were washed using ACN/water (70:30, v/v) and subsequently transferred into either a standard solution (ACN/water, 70:30, v/v) or serum extract⁵ each spiked with MPIX and PPIX (90 pmol each). Loading of porphyrins was performed on a horizontal shaker for 1 h at room temperature. As the pH turned out to be crucial it was adjusted to pH 6 using FA (98–100%, Merck, Darmstadt, Germany) and sodium hydroxide solution (1 M, Merck). For elution of porphyrins the nylon disks were transferred into DMSO (Merck) and shaken for 1 h. A ratio of about 10 μL DMSO to 1 mm² nylon membrane worked best. The extracts were measured in triplicate using our previously developed multiple reaction monitoring (MRM) LC/MS method.⁵

SPE was performed using Resprep MAX (Restek, Bad Homburg, Germany), Oasis MAX (Waters, Eschborn, Germany) and Supel Select Sax (Merck) cartridges; their properties are listed in Table 1. For details on blood and serum extraction refer to our earlier publication.⁵ PPIX purification was carried out either from reference solution (PPIX and MPIX, 50 pmol each, in ACN/water 70:30, v/v; adjusted to pH 8 with aqueous 28–30% ammonia solution) or LLE extracts from whole blood and serum spiked with PPIX and MPIX (50 pmol each) for ae-SPE. Following equilibration of the cartridges with ACN and water the sample was loaded, washed with 5% ammonium hydroxide, methanol and 2% FA in ACN (Figure 3). The eluates were dried using a SpeedVac concentrator (Savant SPD 111 V; Thermo Fisher Scientific, Schwerte, Germany) and reconstituted in 100 μL DMSO for recovery tests and 35 μL DMSO for native serum investigation prior to triplicate LC/MS runs. Sample purification was performed twice per cartridge type. Subsequent experiments omitted sorbent conditioning and equilibration.

原卟啉 (PPIX) 及其前体 5-氨基乙酰丙酸 (ALA) 常规用于胶质母细胞瘤神经外科手术中的荧光引导切除 (FGR)。在手术前口服给药，ALA 优先扩散到携带肿瘤的大脑区域，并在那里转化为 PPIX，作为血红素生物合成的一部分。在高级别胶质瘤 (HGG) 的切除过程中，PPIX 的荧光特性能够区分健康和恶性脑组织，从而实现更完整的肿瘤切除，并最终为患者提供更好的预后。^{1, 2}

尽管使用了 ALA-FGR，但完全切除肿瘤是不可能的——由于 HGG 的浸润性生长，例如多形性胶质母细胞瘤(GBM)，肿瘤经常复发。此外，HGG/GBM 的临床诊断并不简单，需要昂贵的成像技术和组织活检。到目前为止，GBM 患者的预后仍然是现代肿瘤学中最低的之一。^1、3、4

因此，我们研究了 PPIX 作为使用液相色谱/质谱 (LC/MS) 诊断原发性和复发性 GBM 的潜在血液生物标志物。⁵一些使用血浆的癌症研究已经发表。^6-8还证明源自神经胶质瘤细胞的细胞外囊泡含有 PPIX，⁹但这些研究均未使用质谱 (MS) 作为检测器。

全血含有最多的 PPIX，但它不是此类研究的最佳基质，因为红细胞 ZnPPIX 的贡献不明确。然而，对于血清或血浆，用传统方法获得的已发表的 PPIX 水平很低，通常低于检测限，而且不可靠。^{10, 11}利用 MS 的高灵敏度，我们一直致力于为血清的使用提供可重现的 LC/MS 工作流程，并与全血相比测试了我们的方案。⁵因此，从生物基质中可靠地纯化和分别回收 PPIX 被证明是特别困难的，因此我们评估了不同的制备方法。使用水和乙腈 (ACN) 进行的初始液液萃取 (LLE) 对于获得最佳灵敏度至关重要。^5、12然而，需要一个额外的纯化步骤来最大程度地减少 LC 色谱柱的老化和堵塞。

我们首先尝试使用注射器和离心过滤器从大分子中分离出小分子量物质。然而，很快发现，我们测试的一些膜材料——聚四氟乙烯、聚偏二氟乙烯、改性聚醚砜、尼龙和纤维素——特别是尼龙（聚酰胺 6.6）和纤维素——定量吸附了 PPIX。这一观察导致了使用冲孔膜片直接从生物基质中特异性提取 PPIX 的实验，因为随后可以使用二甲基亚砜 (DMSO) 从膜上轻松去除 PPIX。考虑到 PPIX 从参考提取物中的回收率约为 60-70%，粗略初步实验的结果确实很有希望，但处理很困难，由于需要生产和加工小到极小的尼龙碎片，因此可重复性受到了阻碍。卟啉的量与尼龙表面积之间的比率以及上样过程中的 pH 值和洗脱体积至关重要。此外，当使用血清代替标准溶液时，从加标血清中提取中卟啉 (MPIX)（我们的内标）的效率不如 PPIX（图 1）。那时我们改用阴离子交换固相萃取 (ae-SPE)，但我们确实认为应该牢记尼龙在 PPIX 分离方面的潜力。当使用血清而不是标准溶液时，我们来自加标血清的内标不如 PPIX 高效（图 1）。那时我们改用阴离子交换固相萃取 (ae-SPE)，但我们确实认为应该牢记尼龙在 PPIX 分离方面的潜力。当使用血清而不是标准溶液时，我们来自加标血清的内标不如 PPIX 高效（图 1）。那时我们改用阴离子交换固相萃取 (ae-SPE)，但我们确实认为应该牢记尼龙在 PPIX 分离方面的潜力。

最终，ae-SPE 最适合我们的目的，但有趣的是，并非所有吸附剂材料的性能都相同。我们用不同类型的吸附剂测试了三种产品。所有筒床材料均由具有强 ae 性能的季胺官能化聚合物组成；它们仅在聚合物和粒径上有所不同（表 1）。我们评估了一种模拟 LLE 提取物和加标全血以及天然和加标血清的参考溶液。当使用典型的 SPE 工作流程（包括床准备、离子交换、分析物洗涤和洗脱（图 3），从而测定血清中的天然 PPIX未尝试使用此材料。

与硅胶和部分水可湿性材料相比，完全可水润湿的聚合物小柱具有以下优势，即无需执行吸附剂调节和平衡步骤即可直接装载水性样品（图 3）¹³对于亲水-亲油平衡 (HLB) 共聚物，制造商已建议这样做。此外，可以让基于聚合物的墨盒干运行——这在基于真空的处理过程中很容易发生——而不会对性能产生任何负面影响。这简化了ae-SPE 并节省了溶剂和时间。比较所有墨盒类型的这种较短的工作流程（图 2C 和 2D），HLB 共聚物和混合模式 (MM) 吸附剂再次优于基于 Sty 的墨盒类型，前者的整流罩略好于后者。所有这些都没有发现回收率或重现性的损失，也没有发现提取物纯度的差异。使用所有三种试剂盒类型都可以从血清中测定天然 PPIX。MPIX 和 PPIX 三次 LC/MS 运行的平均相对标准偏差 (RSD) 分别为 8% 和 5%，分别。MPIX 的 ae-SPE (n = 2) 的平均 RSD 范围从 HLB 的 3% 到 MMae 小柱的 6%，对于 PPIX，所有产品的 RSD 为 6%。作为张和同事工作的延伸，¹³我们的结果表明，不仅基于 HLB 的吸附剂适用于缩短的 SPE 方案。

另一个决定性因素是使用真空站时的易用性。不建议超过每秒约一滴的流速¹⁴以允许吸附剂和分析物之间的充分相互作用，因为与正相或反相相互作用相比，离子交换过程中的动力学交换过程速度较慢。虽然基于 HLB 和 Sty 的滤芯很容易以推荐的流速运行，但调整 MMae 滤芯的流量既费力又耗时，没有明显的原因。

重要的是，在 ae-SPE 过程中可以正确区分 ZnPPIX 和 PPIX，因为 MPIX 和 PPIX 在 2% 甲酸 (FA) 下洗脱，而 ZnPPIX 没有达到 20% FA 的浓度。⁵考虑到所有参数，我们因此投票支持在我们的研究中使用 HLB 材料。对于从健康志愿者中含有约 2-3 pmol PPIX/mL 的血清 (500 μL) 和全血中提取低丰度的内源性 PPIX，该吸附剂可实现最短的工作流程且易于处理，并且其性能优于 - 即使仅在一种情况下——竞争产品。我们将把我们的程序的使用扩展到肿瘤组织。虽然这里没有展示，但我们希望它在从组织匀浆的 LLE 提取物中纯化 PPIX 时表现同样出色。

实验：使用由尼龙膜（Roth，Karlsruhe，Germany）制成的圆盘进行吸附实验，其表面积为 20 mm ²，用金属空心冲头手动冲出。使用乙腈/水 (70:30, v/v) 清洗圆片，然后转移到标准溶液（乙腈/水，70:30，v/v）或血清提取物⁵中，每个都加入 MPIX 和 PPIX（90 pmol 每个）。卟啉在室温下在水平摇床上加载 1 小时。由于 pH 值被证明是至关重要的，因此使用 FA（98–100%，Merck，Darmstadt，Germany）和氢氧化钠溶液（1 M，Merck）将其调节至 pH 6。为了洗脱卟啉，将尼龙盘转移到 DMSO (Merck) 中并摇动 1 小时。约 10 μL DMSO 与 1 mm 的比例²尼龙膜效果最好。使用我们之前开发的多反应监测 (MRM) LC/MS 方法对提取物进行三次测量。⁵

SPE 使用 Resprep MAX (Restek, Bad Homburg, Germany)、Oasis MAX (Waters, Eschborn, Germany) 和 Supel Select Sax (Merck) 小柱进行；它们的特性列于表 1。有关血液和血清提取的详细信息，请参阅我们早期的出版物。⁵PPIX 纯化是从参考溶液（PPIX 和 MPIX，各 50 pmol，在 ACN/水 70:30，v/v 中；用 28–30% 氨水溶液调节至 pH 8）或全血的 LLE 提取物和加入 PPIX 和 MPIX（各 50 pmol）用于 ae-SPE 的血清。用乙腈和水平衡小柱后，上样样品，用乙腈中的 5% 氢氧化铵、甲醇和 2% FA 洗涤（图 3）。洗脱液使用 SpeedVac 浓缩器（Savant SPD 111 V；Thermo Fisher Scientific，Schwerte，Germany）进行干燥，并在 100 μL DMSO 中进行回收测试，在 35 μL DMSO 中进行天然血清研究，然后进行三次 LC/MS 运行。每个柱类型进行两次样品纯化。随后的实验省略了吸附剂调节和平衡。