Feasibility of PIK3CA Mutation Detection via Microfluidic Isolation of Disseminated Tumor Cells from Bone Marrow in Breast Cancer Patients

Abstract

Disseminated tumor cells originating from the bone marrow serve as precursors of distant metastasis and are associated with increased mortality in breast cancer. With the emergence of targeted therapies for breast cancer, accurate procedures for enrichment of disseminated tumor cells are crucial to investigate the molecular heterogeneity between disseminated tumor cells and primary tumors. This study evaluated the feasibility of PIK3CA mutation analysis in disseminated tumor cells enriched using a microfluidic-based separation system.
Bone marrow and primary tumor samples from 84 breast cancer patients were collected. Disseminated tumor cells were enriched from bone marrow aspirates using the Parsortix system, followed by PIK3CA mutation analysis via MassARRAY PIK3CA Breast Panel and digital droplet PCR.
PIK3CA mutations were detected in 42.9% of primary tumors. PIK3CA mutations were neither detected via digital droplet PCR in disseminated tumor cells-positive bone marrow samples, nor in disseminated tumor cells-negative samples with a high allele frequency of PIK3CA mutations in the primary tumor.
These findings suggest that current microfluidic enrichment methods, such as Parsortix, may be insufficient for reliable PIK3CA mutation detection in disseminated tumor cells. Additional research is required to investigate alternative enrichment techniques for the analysis of mutations in disseminated tumor cells.

Zusammenfassung

Aus dem Knochenmark stammende disseminierte Tumorzellen dienen als Vorläufer von Fernmetastasen und stehen in Zusammenhang mit einer erhöhten Mortalität bei Brustkrebs. Mit dem Entstehen gezielter Therapien für Brustkrebs sind neue Verfahren zur Anreicherung disseminierter Tumorzellen von entscheidender Bedeutung, um die molekulare Heterogenität zwischen disseminierten Tumorzellen und Primärtumoren zu untersuchen. In dieser Studie wurde die Durchführbarkeit der PIK3CA-Mutationsanalyse von disseminierten Tumorzellen getestet, die zuvor mithilfe eines mikrofluiden Systems angereichert wurden, überprüft.
Es wurden Knochenmark- und Primärtumorproben von 84 Brustkrebspatientinnen entnommen. Die disseminierten Tumorzellen wurden aus Knochenmarkaspiraten mit dem Parsortix-System angereichert, gefolgt von einer PIK3CA-Mutationsanalyse mittels MassARRAY PIK3CA Breast Panel und Digital Droplet-PCR.
PIK3CA-Mutationen wurden in 42,9% der Primärtumoren nachgewiesen. PIK3CA-Mutationen wurden weder mittels Digital Droplet-PCR in disseminierten tumorzellpositiven Knochenmarkproben noch in disseminierten tumorzellnegativen Proben mit einer hohen Allelfrequenz von PIK3CA-Mutationen im Primärtumor nachgewiesen.
Diese Ergebnisse deuten darauf hin, dass aktuelle mikrofluide Anreicherungsmethoden wie Parsortix möglicherweise keinen zuverlässigen Nachweis von PIK3CA-Mutationen in disseminierten Tumorzellen ermöglichen. Es sind weitere Studien erforderlich, um alternative Anreicherungsverfahren für die Mutationsanalyse in disseminierten Tumorzellen zu erforschen.

Keywords

Parsortix - disseminated tumor cells (DTCs) - breast cancer - liquid biopsy - digital droplet PCR (ddPCR)

Schlüsselwörter

Parsortix - disseminierte Tumorzellen (DTCs) - Brustkrebs - Liquid Biopsy - Digital Droplet-PCR (ddPCR)

Introduction

Breast cancer (BC) remains a substantial global health challenge, with metastasis serving as the primary determinant of mortality in affected patients [1] [2]. In this context, micro-metastasis in the form of disseminated tumor cells (DTCs) have been identified as elements in the metastatic cascade [3]. DTCs are found in approximately 30% of breast cancer patients at the time of primary diagnosis [4] [5] [6] [7]. These cells are a rare population, presenting as either single cells or cell clusters [8]. Regardless of other factors, the presence of DTCs in the bone marrow (BM) is predictive for a poorer overall survival rate [9] [10], being associated with adverse prognostic factors such as advanced tumor stage, poor differentiation, lymph node metastasis [6] [11] and disease recurrence [12].

DTCs pose a clinical challenge due to their ability to evade conventional therapies by entering a state of dormancy, potentially leading to late recurrence [5] [8] [13]. These cells can disseminate to distant sites at a later point and may establish distant metastases, predominately in the bone, occasionally to the liver, lung and brain [8]. Studies suggest that dissemination of single tumor cells has already taken place in the majority of patients by the time of primary breast cancer detection [13].

Currently, BM is most commonly obtained during primary breast cancer surgery and DTCs are detected using immunohistochemistry stainings targeting epithelial breast cancer markers such as EpCAM or cytokeratin. This method is most widely used and clinically established for prognostic assessment [14]. Other enrichment methods are also antibody-based, for example using EpCAM and HER2 as targets [15] [16]. It has been noted that circulating tumor cells (CTCs) and DTCs may exhibit different biological profiles compared to the primary tumor (PT) [17] [18] and may even loose the expression of some epitopes. This is particularly relevant in the context of Epithelial-to-Mesenchymal Transition (EMT), as this process induces significant changes in cell surface markers and can lead to the downregulation or loss of epithelial epitopes while promoting the expression of mesenchymal markers [19]. Therefore, immunohistochemical detection of DTCs is challenged by potential underdiagnosis and the possibility to yield only a limited number of cells available for analysis.

Previous challenges in isolating, identifying, and analyzing DTCs due to their heterogeneity and the limited amount of cells have been addressed with microfluidic filtration for epitope-independent enrichment and isolation of DTCs such as the Parsortix method [18] [20] [21]. This system relies on physical properties such as cell size, decreased deformability and cell density [22]. A previous study from our group validated the isolation of DTCs using Parsortix which yielded significantly higher cell counts than the standard immunohistochemistry method [18]. This leads to an expanded analytical capacity and could open new diagnostic possibilities, such as the detection of mutations with therapeutic relevance in DTCs.

Approximately 30% of BC cases, particularly hormone receptor-positive (HR+) and HER2-negative tumors, manifest mutations within the PI3K pathway [23] [24]. Hotspot mutations of the PIK3CA gene, that dysregulate the PI3K pathway, are frequently observed in BC patients [25] [26]. Their clinical significance has been substantiated, as PI3K inhibitors, such as alpelisib or inavolisib, are effective as a targeted therapy in metastatic HR+/HER2− breast cancer with PIK3CA mutations [27] [28]. Moreover, since PIK3CA mutations lead to downstream signaling through AKT, AKT inhibitors like capivasertib have demonstrated significantly longer median progression-free survival in PIK3CA-mutated, HR+ advanced breast cancer [29]. Discrepancies regarding the PIK3CA mutation status between CTCs, DTCs and metastases have been discovered in previous studies [8] [30] [31], emphasizing the importance of considering the PIK3CA status in both PT and micro-metastases for therapeutic decisions in early breast cancer [32].

As a detection method for mutations, digital droplet polymerase chain reaction (ddPCR) allows precise and absolute quantification of target molecules by partitioning samples into numerous droplets, which is beneficial for detecting mutations in rare cells like DTCs [33]. Compared to standard PCR, ddPCR offers enhanced sensitivity, specificity and precision for detecting mutations down to 0.1% detection limit, which enables the accurate quantification of low DNA amounts, such as in DTCs [34].

There is a need to investigate DTCs in BC at the molecular level to understand their role in metastasis and their potential as therapeutic targets. The aim of this study was to evaluate an improved clinical procedure for DTC enrichment and for the detection of critical hotspot mutations such as PIK3CA.

Material and Methods

Ethic statement

The patient BM samples originate from the University Women’s Hospital in Tübingen as part of the primary surgical procedure of BC patients. The study was conducted in accordance with protocols approved by the Ethics Committee of the Eberhard Karls University of Tübingen (reference numbers 528/2019BO2 and 040/2023BO2) and all relevant ethical regulations for research with human participants were observed. The patients were informed and gave their consent to the scientific use of their BM samples.

Patient cohort

Patients with the following criteria regarding the probability of a PIK3CA mutation were included in the cohort: diagnosis of invasive breast cancer (NST/ILC/other), no distant metastases detected at time of first BM aspiration, positivity in estrogen receptor (ER +) and progesterone receptor (PR +), HER2 status negative/low, cryopreserved BM samples. Patients who received neoadjuvant therapy were excluded. BM was collected and cryopreserved from 84 BC patients meeting inclusion criteria for this analysis between October 2020 and December 2022 (see online Supplementary Fig. S1 for patient and sample selection).

Cell culture, reagents and fundamentals

MCF7 and SK-BR-3 cell lines used in the present study were obtained from ATCC. The DMEM (Dulbecco’s modified eagle medium) supplemented with 10% FBS (fetal bovine serum), 50 μg/mL penicillin/streptomycin, and 2 mM L-glutamine (all reagents were purchased from Thermo Fisher Scientific, Waltham, MA, USA) were used for cell cultivation. Cells were incubated at 37 °C in a humidified atmosphere containing 5% CO₂. For routine passaging, cells were washed with 1× PBS (Dulbecco’s phosphate-buffered saline, Thermo Fisher Scientific, MA, USA) and treated with Trypsin-EDTA 0.05% (Sigma Aldrich, St. Louis, MO, USA) as recommended by ATCC product sheets.

Bone marrow preparation

The samples were taken via a BM puncture at the anterior superior spina iliac crest under sterile conditions in the operating room under general anesthesia and filled into 20 mL syringes (mixed with 1000 IE heparin).

BM samples were mixed with PBS (ratio 1 : 3) to achieve a viscosity and volume similar to that of blood. BM was filtered through a 100 μm cell strainer for removal of bone trabeculae and then centrifuged at 478× g for 10 minutes. The supernatant was discarded after centrifugation and the cell pellet was resuspended in 10 mL PBS + BSA 1% and transferred to a VacuTainer tube containing 1.8 mg/mL K2EDTA (Becton Dickinson, New York, NY, USA).

Cryopreservation and thawing procedure of bone marrow cells

For cryopreservation the BM preparation was performed as described above and an additional 10 min red blood cell lysis (RBC) at 4 °C was performed using Red Blood Cell Lysis buffer (sterile prepared, buffer: 155 mM NH4Cl + 10 mM KHCO3 + 100 μM Na2EDTA in H₂O; pH: 7.4). This was followed by another centrifugation at 478× g for 10 minutes with brake, so that after aspiration of the supernatant a cell pellet with the mononuclear cells (including tumor cells) remained. The resulting cell pellet was then resuspended with freezing medium (Recovery Cell Culture Freezing Medium, Thermo Fisher Scientific, Fremont, USA) and cryopreserved in cryovials at −80 °C freezer, placed 4 h in freezing containers (CoolCell LX, Corning, NY, USA) at the rate of −1 °C/minute before long-term storage in liquid nitrogen (−196 °C).

In order to analyze the samples retrospectively, the cryopreserved samples were thawed by immersing the cryovials in a water bath at 37 °C for 45 seconds. 100 μL DNase I (Thermo Fisher Scientific, MA, USA) solution, followed by 1000 μL medium (DMEM + 10% FBS), both pre-cooled at 4 °C, were added slowly, drop by drop. The cell suspension was taken up with a BSA coated pipette tip and transferred slowly into a tube containing 15 mL of pre-cooled medium (DMEM + 10% FBS). For the homogenization, the tube was carefully rolled three times. The sample was centrifuged at 300× g for 10 minutes. The cell pellet containing the tumor cells were resuspended in PBS + 0.1% BSA and transferred to a VacuTainer tube containing 1.8 mg/mL K2EDTA.

Microfluidic cell separation

The Parsortix PR1 automated microfluidic system (Angle North America, Philadelphia, PA, USA) was used to isolate the DTCs. This microfluidic cell separation platform was previously described by Pillai et al. [18]. The manufacturer’s protocol scheme including weekly full cleans (PX_C) and maintenance cleans.

According to the manufacturer’s instructions, the 6.5 μm gap size separation cassette (ANGLE PLC, Surrey, UK) was used, for optimal number of collected DTCs and lowest number of PBMCs in the cell harvest. The cassette was primed with 100% ethanol, followed by PBS. The process of cell separation using the manufacturer’s manual of Parsortix contains the priming PX2_F, the separation PX2_S99F, and the harvest PX2_H. The cells were harvested in a two-stage process with first a 200 μL PBS harvest and a following second 1000 μL PBS reverse flow. After every sample the Parsortix was cleaned using the PX2_C protocol according to the manufacturer’s protocol.

The estimated number of DTCs to be expected (number of expected cells after Parsortix cell separation = number of total numbers of initially present DTCs (evaluated during BM processing) × harvest capacity (0.6×) of Parsortix method) were 159 and 114.

Spiking experiments

Spiking experiments were conducted to validate the harvest rate of Parsortix using the SK-BR-3 cell line as breast cancer research model. [Fig. 1] illustrates the setup of the spiking experiments. Approximately 200 SK-BR-3 cells were spiked into fresh BM. The exact number of spiked cells was quantified using the AutomatedCellCounter TC20 (Biorad Laboratories, Inc.), followed by direct cell counting on two microscopic slides for validation. To facilitate cell counting, Cell Tracker Green fluorescent staining was employed. Cell separation was performed using Parsortix. Capture and harvest rates were calculated by enumerating cells in the 6.5 μm cassette before and after harvest under a fluorescent microscope. Following separation, harvested cells were collected on a cover slip, subjected to cytospin centrifugation, fixed with formaldehyde (3.7%) using Prolong diamond DAPI as mounting medium. Enumeration of positive cells stained with CellTracker Green and DAPI was performed using an EVOS M7000 microscope (Thermo Fisher Scientific, MA, USA) at 10× and 20× magnification. Capture rate was defined as the ratio of spiked cells to captured cells in the cassette, while harvest rate represented the ratio of captured cells to harvested cells. Recovery rate was calculated as the ratio of spiked cells to harvested cells.

Fig. 1 Workflow of spiking experiments with Parsortix, BM = bone marrow.

To validate the analytical sensitivity of digital droplet PCR, specific amounts of MCF-7 breast cancer cells were spiked into fresh BM. MCF7 were chosen because the cell line is mutated in exon 9 (E545K hotspot mutation) [35]. Cell separation was performed using Parsortix, and DNA isolation was carried out using the Zymo Research Quick-DNA/RNA Miniprep Kit. Spiked cells in various concentrations of peripheral blood mononuclear cells (PBMCs) served as a control.

Concentration of cells in BM was determined using Sysmex hemocytometer (XP-300 Automated Hematology Analyzer, Sysmex Europe SE, Norderstedt, Germany).

In order to quantify the impact of PBMC contamination on the detection of PIK3Ca mutations, the percentage change in FA in relation to PBMC concentrations were calculated.

Isolation of DNA from FFPE tissue

The tumor area was identified by pathologists on hematoxylin-eosin (HE) stained slides, guiding the selection of tumor tissue from the FFPE block for DNA isolation. For all included patients, the tumor areal was precisely defined, and the FFPE tissue material used depending on the tumor size.

Paraffin sections of formalin-fixed paraffin-embedded (FFPE) tissue were used for DNA extraction from primary breast cancer tumors. Four sections were cut per tissue block using a microtome set to a thickness of 10 μm. Hematoxylin-eosin (HE) stainings were used to delineate the tumor area. Paraffin shavings containing the selected tumor area were collected and transferred to a 1.5 mL Eppendorf tube using a scalpel and forceps for DNA isolation. The tissue was either spread on microscope slides or directly transferred into Eppendorf tubes. After each sample, the scalpel and forceps were cleaned with a xylene substitute.

DNA was isolated using the Quick DNA/RNA FFPE kit according to the manufacturer’s instructions. The protocol included steps for buffer preparation, sample preparation, DNA and RNA purification and DNase I treatment for RNA purification. The concentrations and purity of the DNA and RNA samples were determined using a VarioScan spectrophotometer (Thermo Fisher Scientific, MA, USA) by measuring absorbance at 260 nm and 280 nm.

Genomic profiling for the detection of PIK3CA hotspot mutation

The isolation of DNA from the harvest of the microfluidic cell separation was performed by the Zymo Research Quick-DNA/RNA Miniprep Kit.

Primary tumors and the BM samples were first analyzed using the PIK3CA Breast Panel on the MassARRAY System for all clinically relevant PIK3CA mutations at 1% sensitivity (Agena Bioscience). For quality control, positive samples including those with borderline signals from the MassARRAY analysis were re-tested for four mutations (E542K c.1624G>A, E545K c.1633G>A, H1047R c.3140A>G, N345K c.1035T>A) using digital droplet ddPCR at 0.1% sensitivity (Biorad system). PCR products were loaded onto the QX200 Droplet Reader (Bio-Rad Laboratories, Inc.) and the analysis of the data was performed using QXManager. Variant allele frequency (VAF) was calculated as the proportion of variant to wildtype signal. The test results were negative (below threshold on the MassARRAY or 0 positive droplets, weakly positive (2–20 positive droplets in ddPCR or VAF < 0.5%) or strongly positive (> 20 droplets in ddPCR or VAF > 1% on the MassARRAY). Two patients had double mutations, which are counted once in the total number of patients but contribute twice to the mutation counts.

Imaging and visualization

Imaging was performed using EVOS M7000 Imaging system (Thermo Fisher Scientific, MA, USA) and ScanScope imaging system (Leica Biosystems, DE).

Statistical analysis

All data was recorded in a Microsoft Excel spreadsheet (Version 2010, Microsoft Corporation, Redmond, Washington DC, USA). The statistical analysis was performed using SPSS 15.0 (SPSS Inc., Chicago, IL, USA). The correlation between clinical and pathological parameters and the detection of DTCs in the BM as well as PIK3CA mutation was evaluated using the Chi-Square test.

Results

Validation of microfluidic separation as a means to isolate DTCs

The Parsortix microfluidic-based system was used to enrich DTCs. To assess the effectiveness and sensitivity of DTC isolation via Parsortix, we conducted a spiking experiment using a common breast cancer cell line. For this purpose, fresh BM samples were spiked with CellTracker Green labelled cells, followed by standard density centrifugation, then processed through microfluidic cell separation, and captured by cytospin ([Fig. 1] and [Fig. 2]). Our results showed a capture rate of 82.76%, a harvest rate of 50%, and a recovery rate of 41.38% (Supplementary Table S1 online). These results validate the ability of isolating DTCs in high numbers via microfluidic separation and the harvest rate from our previous study.

Fig. 2 Immunofluorescence images of harvested cells after Parsortix isolation of DTCs from BM. Celltracker Green was captured by GFP mode and merged with DAPI, scalebar 10 µm.

Validation of ddPCR for detection of PIK3CA hotspot mutations in spiked MCF7 cells

To investigate the detection of PIK3CA positive DTCs by ddPCR, we spiked MCF7 cells (PIK3CA mutated, E545K) into fresh BM. Cell separation was performed using Parsortix. The presence of the PIK3CA mutation was successfully identified utilizing ddPCR in all samples containing 500, 800, and 20000 cells (see [Table 1]). The assay demonstrated a lower detection limit for the PIK3CA mutation at 160 spiked MCF7 cells per sample, corresponding to approximately 230.00 BM cells when considering a sensitivity of 0.14% and a recovery rate of 50%.

Table 1 Results of the detection limits for the PIK3CA mutation in ddPCR.

Number of MCF7 cells	Fractional abundance of PIK3CA mutation (%)	Rated result
20000	7.36	Positive
800	0.46	Positive
500	0.51	Positive
160	0.14	Borderline positive (1 droplet)

Study cohort

To determine the feasibility of routine PIK3CA screening in DTCs at a clinical level, BM samples of patients with early BC were analyzed. BM samples were obtained during primary surgical procedure, and subsequently processed and cryopreserved. Patients with clinical and pathological parameters related to a higher probability of a PIK3CA mutation (ER+, PR+ hormone receptor status, HER2 negativity) were included in this cohort. Patients treated with any neoadjuvant therapy or endocrine therapy were excluded from this study.

A total of 84 BC patients were included into the analysis (see [Table 2] for clinical and histopathological details). Most patients were postmenopausal (n = 56, 67%) and the dominant tumor type was no-special type (NST, n = 52, 62%). The majority had a pathological tumor size of pT1–2 (n = 47, 56%) and showed no involvement of lymph nodes (n = 47, 56%). The median age was 60. Cryopreserved BM was available for all 84 patients. Among these samples, ten were classified as DTC-positive using standard detection method of cytokeratin staining. In this cohort, DTC-positivity was not significantly associated with the classical histopathological parameters.

Table 2 Distribution of DTC positivity across clinical and pathological parameters.

Parameter	n	DTC positivity, n (%)	P value*
ILC= invasive lobular carcinoma; NST= no-special type * P values were calculated to assess the significance of differences in DTC positivity across subgroups using Pearson’s chi-square test
Total	84	10/84 (12%)
Histology			0.728
NST	60	8/60 (13%)
ILC	21	2/21 (9%)
Other	3	0/3 (0%)
Tumor size			0.409
pT1–2	52	5/52 (10%)
pT3–4	32	5/32 (16%)
Nodal status			0.900
N0	53	6/53 (11%)
N+	30	4/30 (13%)
Menopausal status			0.579
Premenopausal	17	1/17 (6%)
Postmenopausal	67	9/67 (13%)

Detection of PIK3CA mutations in primary tumors

For all 84 patients, PTs were analyzed for the presence of PIK3CA hotspot mutations. Overall, PIK3CA mutations were detected in 36 out of 84 PT (43%). When stratified by histopathological parameters (see [Table 3]), there was no statistically significant difference in PIK3CA positivity. However, PIK3CA mutations were more frequently detected in premenopausal patients (71% vs. 36% in postmenopausal patients, p = 0.021). The analysis focused on known hotspot mutations within the PIK3CA gene, particularly those at codons 542, 545 and 1047, which are frequently associated with oncogenic activity. Among these, 42.9% (36 patients, two patients double mutated) were tested positive for PIK3CA mutation. The most frequently observed mutation was H1047R, detected in 16 patients (42.1%), see [Fig. 3].

Table 3 Distribution of PIK3CA positivity found in ddPCR across clinical and pathological parameters.

Parameter	n	PIK3CA positivity, n (%)	P value*
ILC = invasive lobular carcinoma; NST = no-special type * P values were calculated to assess the significance of differences in DTC positivity across subgroups using Pearson’s chi-square test
Total	84	36/84 (43%)
Histology			0.580
NST	60	24/60 (40%)
ILC	21	10/21 (5%)
Other	3	2/3 (67%)
Tumor size			0.559
pT1–2	52	21/52 (40%)
pT3–4	32	15/32 (47%)
Nodal status			0.283
N0	53	20/53 (38%)
N+	30	15/30 (50%)
Menopausal status			0.021
Premenopausal	17	12/17 (71%)
Postmenopausal	67	24/67 (36%)

Fig. 3 Distribution of PIK3CA mutations in primary breast tumors. The PIK3CA gene is shown schematically along the x-axis (0–1068 amino acids), with hotspot mutations indicated by color-coded bars at their codon positions. Vertical lines represent the number of patients per detected mutation.

Of the 36 patients who were positive for the PIK3CA mutation in the PT, two (5.6%) were also found to have DTCs in their BM, when using cytokeratin immunohistochemistry staining as the standard method for DTC detection.

Evaluation of corresponding PIK3CA mutations detected by ddPCR in disseminated tumor cells

Based on the results of the spiking experiments, which demonstrated that DTC isolation and subsequent PIK3CA mutation detection were feasible, we analyzed a total of four BM samples from BC patients as a proof-of-principle test for its clinical use. The BM samples were examined for PIK3CA mutations under the assumption that the DTCs present in these samples carry this mutation. From the 36 PT which have been tested positive for PIK3CA mutations, two BM samples that showed DTCs by standard method were evaluated. Based on the previous spiking experiments, an estimated 159 and 114 DTCs would be isolated via Parsortix in these two BM samples (Supplementary Table S2 online).

Another two BM samples with a notably high allele frequency of the PIK3CA mutation in the PT while being DTC-negative by the standard method were also evaluated.

All four samples were subjected to ddPCR analysis to test for the presence of PIK3CA mutations found in the PT. For all four BM samples, none of the PIK3CA hotspot mutations present in the PT (E545K c.1633G>A, H1047R c. 3140A>G) were identified via the ddPCR assay.

Impact of PBMC contamination on PIK3CA mutation detection in ddPCR

Since the PIK3CA hotspot mutations present in the PT were not identified via ddPCR in the patient’s BM samples, we further investigated potential challenges for PIK3CA testing. To this end, we investigated the effects of PBMCs contamination in the final harvest after DTC isolation via Parsortix. We performed spiking experiments by introducing a defined number of MCF7 cells (100 or 500) into samples containing varying quantities of PBMCs (20000, 100000 or 500000). The results demonstrate a correlation between the extent of PBMC contamination and the fractional abundance (FA) of PIK3CA mutation detection in ddPCR ([Table 4]). For instance, in spike-in samples with 100 MCF7-cells increasing PBMCs from 20000 to 100000 resulted in a decrease in FA, indicating a substantial reduction in mutation detection sensitivity close to the lower detection limit at higher PBMC contamination. Altogether our data show that the FA at higher PBMC concentration is only at 4%–12% compared to the FA at a lower PBMC concentration.

Table 4 Fractional abundance of detected PIK3CA mutation via ddPCR in spiking experiments of MCF7 cells in PBMC.

Number of MCF7 cells spiked into PBMCs	Number of PBMCs	PIK3CA mutation fractional abundance FA (%)	Rated result	% of FA vs. low PBMC*
* % of FA vs. Low PBMC: Fractional abundance relative to the sample with the lowest PBMC contamination within the same MCF7 spike-in group (set as 100%).
100	20000	1.81	Positive	100
100	100000	0.22	Borderline positive	12.2
500	20000	4.90	Positive	100
500	100000	0.55	Positive	11.2
500	500000	0.18	Borderline positive	3.7

Discussion

The characterization of DTCs is relevant for understanding BC progression and its impact in patient outcomes [11]. Studies have shown that DTCs in BM are associated with disease recurrence even after the completion of standard therapy [12] [36], highlighting their clinical significance in long term patient prognosis. The understanding of the biology, behavior, and clinical implications of DTCs in BC is therefore paramount for the development of prognostic markers and potentially also for targeted therapies. However, the low number of cells and limitations of current detection methods pose a major challenge in the molecular characterization of DTCs.

Our recent investigation with microfluidic separation of DTCs has shown a higher percentage of harvested cells compared to the standard method of immunohistochemistry staining [18]. These results are in line with previous studies that have demonstrated variable yields in DTC isolation from BM depending on different methodologies. In this study, a novel approach was employed by combining DTC isolation with mutational genotyping [18]. Spiking experiments were conducted to simulate the presence of DTCs in BM and evaluate the performance of the Parsortix system in isolating and detecting these cells. Our experiments demonstrated a capture rate of 82.8% and a recovery rate of 41.4%. These results exceed the 65,6% capture rate previously reported for Parsortix-based DTC isolation from BM [18] and align with DTC recovery rate studies ranging between 26% in DTC isolation from BM [20] to 54–69% for CTCs and 72.8% for melanoma DTCs [22] [37] [38]. Differences in capture and recovery rates across studies may originate from variations in device settings, sample processing, and intrinsic tumor cell properties. Nevertheless, our data further emphasize the potential of microfluidic approaches and the reliability of detecting PIK3CA mutations.

PIK3CA mutations are recognized as one of the most prevalent genetic alterations in BC, with implications for both tumor biology and clinical management [39]. Targeted therapies aimed at inhibiting the PI3K/AKT/mTOR pathway are established therapeutic options for BC patients with PI3K-mutant tumors [27] [40]. In this study, the analysis of primary BC tumors revealed a distribution of PIK3CA mutations similar to previous studies [25] [26] [39] [41]. However, PIK3CA mutations were not detected in the analyzed BM samples containing DTCs. Potential reasons for the absence of PIK3CA mutations in ddPCR analysis with state-of-the-art sensitivity of 0.1% include both tumor heterogeneity and technical limitations. Given the molecular heterogeneity between DTCs, macrometastasis and PT [17] [42], and the evolution of genotypic divergence over time [43], it may be possible that DTCs in the BM samples were truly PIK3CA negative, reflecting biological tumor heterogeneity. In contrast, several studies report a high concordance of PIK3CA mutation status between primary tumors and distant metastases in BC, which argues against extensive genotypic divergence during metastatic spread [44] [45]. In addition, the limited sample size in our study with two samples both PIK3CA-mutated in the PT and DTC-positive further restricts the interpretive strength of this assumption and reduces the probability that the DTCs were genuinely PIK3CA-negative due to biological heterogeneity alone.

However, technical limitations must be considered as well. The extremely low number of DTCs typically present in BM samples, with reported rates down to only 10–20 cancer cells per million BM cells [11], holds the chance of false-negative results due to limited detection sensitivity. Our spike-in sensitivity experiments indicate a detection limit for PIK3CA mutations of approximately 160 cells per BM sample using a combined microfluidic separation and ddPCR detection methodology. Among several technical limitations, the impact of cryopreservation should be considered, as it may lead to cell loss, further compromising detection sensitivity. In this study, however, cryopreservation was necessary due to the retrospective analysis of the histopathological details.

In addition, contamination with non-target cells, particularly mononuclear and BM stem cells, poses a major challenge. We suspect that the microfluidic cell separation method may have contributed to an overrepresentation of these non-tumor mononuclear and BM stem cells in the isolated samples, thereby diluting the DTC population and potentially obscuring the PIK3CA signal. This hypothesis is supported by our spiking experiments, which evaluated the detection limit for mutations in small quantities of DTCs in the presence of low and high numbers of PBMCs using ddPCR. In this context, a challenge of the Parsortix method remains its tendency to co-isolate mononuclear BM cells, as previously described [20]. Thus, depletion steps may be necessary to ensure higher purity of isolated DTCs. Altogether, this highlights the necessity of optimizing DTC-isolation methods to improve mutation analysis in BM samples. An alternative approach to improve DTC detection and characterization could be the use of tumor-agnostic single-cell sequencing to detect driver gene mutations or cancer-specific chromosomal copy number changes, which however has limited use for routine clinical practice.

The present study evaluated the Parsortix method followed by downstream ddPCR mutation analysis of DTCs. Our findings suggest that Parsortix is neither controllable nor predictable in terms of the quantity of non-target cells. Consequently, the degree of cell contamination during DTC-isolation and the number of DTCs recovered cannot be accurately assessed using this method. Microfluidic isolation of DTCs via Parsortix might be a good option for the molecular analysis of DTCs under specific conditions, such as higher DTC counts. However, its limitations underscore the need for more sensitive and reliable enrichment techniques for detecting mutations in DTCs in clinical routine.

Conclusion

By using a combination of microfluidic detection and tumor-informed mutation analysis of DTCs, our study provides valuable insights into the technical and biological challenges of characterizing DTCs in BC. The microfluidic Parsortix method leads to harvest of much higher cell numbers than in conventional methods. However, our findings highlight limitations in its application to downstream molecular analysis, due to unpredictable contamination with non-target cells. Although the method may be applicable in settings with higher DTC counts, its current version may not be suitable for routine clinical implementation. There is a need for more sensitive enrichment methods that ensure both high DTC purity and compatibility with molecular assays.

Supplementary Material

• Supplementary Fig. S1: Flowchart of patient and sample selection.

• Supplementary Table S1: Summary of defined metrics in Spiking Experiment following Parsortix-based DTC isolation. Cell enumeration was performed via immunofluorescence imaging and cytospin analysis.

• Supplementary Table S2: Estimated numbers of DTCs after Parsortix Enrichment in BM samples from PIK3CA positive patients.

Conflict of Interest

AH reported honoraria, consulting roles, travel support and research support from: Roche, Novartis, Lilly, MSD, AstraZeneca, Daiichi Sankyo, Seagen, GSK, ExactScience, Gilead, Menarini Stemline, Pfizer, Eisai, Veracyte, Agendia, Riemser, Teva, Onkowissen, Amgen, Pierre Fabre, Thieme, Springer. The other authors have no relevant financial or nonfinancial interests to disclose.

Acknowledgement

We thank Ingrid Teufel, Susanne Bruckner, Angelika Amann, Sabine Hofmeister and Anna Kechter for their excellent technical assistance. We further acknowledge Lisa Meisl, Lena Mesch and Juliane Köstlin for their support in proofreading and for valuable comments on the manuscript. Finally, we thank all patients for their participation in this study.

• References

• 1 Robert Koch-Institut. Krebs in Deutschland für 2019/2020. 2023
• 2 Bray F, Laversanne M, Sung H. et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024; 74: 229-263
• 3 Muscarella AM, Aguirre S, Hao X. et al. Exploiting bone niches: progression of disseminated tumor cells to metastasis. J Clin Invest 2021; 131: e143764
• 4 Riethdorf S, Wikman H, Pantel K. Review: Biological relevance of disseminated tumor cells in cancer patients. Int J Cancer 2008; 123: 1991-2006
• 5 Banys M, Hartkopf AD, Krawczyk N. et al. Dormancy in breast cancer. Breast Cancer (Dove Med Press) 2012; 4: 183-191
• 6 Braun S, Vogl FD, Naume B. et al. A Pooled Analysis of Bone Marrow Micrometastasis in Breast Cancer. N Engl J Med 2005; 353: 793-802
• 7 Hartkopf AD, Brucker SY, Taran FA. et al. Disseminated tumour cells from the bone marrow of early breast cancer patients: Results from an international pooled analysis. Eur J Cancer 2021; 154: 128-137
• 8 Ring A, Spataro M, Wicki A. et al. Clinical and Biological Aspects of Disseminated Tumor Cells and Dormancy in Breast Cancer. Front Cell Dev Biol 2022; 10: 929893
• 9 Wiedswang G, Borgen E, Kåresen R. et al. Detection of Isolated Tumor Cells in Bone Marrow Is an Independent Prognostic Factor in Breast Cancer. J Clin Oncol 2003; 21: 3469-3478
• 10 Hartkopf AD, Brucker SY, Taran FA. et al. Disseminated tumour cells from the bone marrow of early breast cancer patients: Results from an international pooled analysis. Eur J Cancer 2021; 154: 128-137
• 11 Riethdorf S, Pantel K. Disseminated Tumor Cells in Bone Marrow and Circulating Tumor Cells in Blood of Breast Cancer Patients: Current State of Detection and Characterization. Pathobiology 2008; 75: 140-148
• 12 Gruber I, Fehm T, Taran FA. et al. Disseminated tumor cells as a monitoring tool for adjuvant therapy in patients with primary breast cancer. Breast Cancer Res Treat 2014; 144: 353-360
• 13 Friberg S, Nystrom A. Cancer Metastases: Early Dissemination and Late Recurrences. Cancer Growth Metastasis 2015; 8: CGM.S31244
• 14 Fehm T, Braun S, Muller V. et al. A concept for the standardized detection of disseminated tumor cells in bone marrow from patients with primary breast cancer and its clinical implementation. Cancer 2006; 107: 885-892
• 15 Schindlbeck C, Andergassen U, Jueckstock J. et al. Disseminated and circulating tumor cells in bone marrow and blood of breast cancer patients: properties, enrichment, and potential targets. J Cancer Res Clin Oncol 2016; 142: 1883-1895
• 16 Cackowski FC, Wang Y, Decker JT. et al. Detection and isolation of disseminated tumor cells in bone marrow of patients with clinically localized prostate cancer. Prostate 2019; 79: 1715-1727
• 17 Magbanua MJM, Rugo HS, Hauranieh L. et al. Genomic and expression profiling reveal molecular heterogeneity of disseminated tumor cells in bone marrow of early breast cancer. NPJ Breast Cancer 2018; 4: 31
• 18 Volmer LL, Önder CE, Volz B. et al. Microfluidic Isolation of Disseminated Tumor Cells from the Bone Marrow of Breast Cancer Patients. Int J Mol Sci 2023; 24: 13930
• 19 Alix-Panabières C, Mader S, Pantel K. Epithelial-mesenchymal plasticity in circulating tumor cells. J Mol Med (Berl) 2017; 95: 133-142
• 20 Pillai SG, Siddappa CM, Ma C. et al. A microfluidic-based filtration system to enrich for bone marrow disseminated tumor cells from breast cancer patients. PLoS One 2021; 16: e0246139
• 21 Miller MC, Robinson PS, Wagner C. et al. The ParsortixTM™ Cell Separation System—A versatile liquid biopsy platform. Cytometry A 2018; 93: 1234-1239
• 22 Hvichia GE, Parveen Z, Wagner C. et al. A novel microfluidic platform for size and deformability based separation and the subsequent molecular characterization of viable circulating tumor cells. Int J Cancer 2016; 138: 2894-2904
• 23 Mosele F, Stefanovska B, Lusque A. et al. Outcome and molecular landscape of patients with PIK3CA-mutated metastatic breast cancer. Ann Oncol 2020; 31: 377-386
• 24 Nunnery SE, Mayer IA. Targeting the PI3K/AKT/mTOR Pathway in Hormone-Positive Breast Cancer. Drugs 2020; 80: 1685-1697
• 25 Gonzalez-Angulo AM, Ferrer-Lozano J, Stemke-Hale K. et al. PI3K Pathway Mutations and PTEN Levels in Primary and Metastatic Breast Cancer. Mol Cancer Ther 2011; 10: 1093-1101
• 26 Martínez-Sáez O, Chic N, Pascual T. et al. Frequency and spectrum of PIK3CA somatic mutations in breast cancer. Breast Cancer Res 2020; 22: 45
• 27 Fusco N, Malapelle U, Fassan M. et al. PIK3CA Mutations as a Molecular Target for Hormone Receptor-Positive, HER2-Negative Metastatic Breast Cancer. Front Oncol 2021; 11: 644737
• 28 Turner NC, Im S-A, Saura C. et al. Inavolisib-Based Therapy in PIK3CA-Mutated Advanced Breast Cancer. N Engl J Med 2024; 391: 1584-1596
• 29 Würstlein R, Kolberg H-C, Hartkopf AD. et al. Update Breast Cancer 2024 Part 1 – Expert Opinion on Advanced Breast Cancer. Geburtshilfe Frauenheilkd 2024; 84: 529-540
• 30 Schmidt-Kittler O, Ragg T, Daskalakis A. et al. From latent disseminated cells to overt metastasis: genetic analysis of systemic breast cancer progression. Proc Natl Acad Sci U S A 2003; 100: 7737-7742
• 31 Aguirre-Ghiso JA. Models, mechanisms and clinical evidence for cancer dormancy. Nat Rev Cancer 2007; 7: 834-846
• 32 Dupont Jensen J, Laenkholm A-V, Knoop A. et al. PIK3CA Mutations May Be Discordant between Primary and Corresponding Metastatic Disease in Breast Cancer. Clin Cancer Res 2011; 17: 667-677
• 33 Luthra R, Singh RR, Patel KP. , ed. Clinical Applications of PCR. 3. New York, Heidelberg: Humana Press; 2016
• 34 Postel M, Roosen A, Laurent-Puig P. et al. Droplet-based digital PCR and next generation sequencing for monitoring circulating tumor DNA: a cancer diagnostic perspective. Expert Rev Mol Diagn 2018; 18: 7-17
• 35 Hollestelle A, Elstrodt F, Nagel JHA. et al. Phosphatidylinositol-3-OH Kinase or RAS Pathway Mutations in Human Breast Cancer Cell Lines. Mol Cancer Res 2007; 5: 195-201
• 36 Hartkopf AD, Taran F-A, Wallwiener M. et al. Prognostic relevance of disseminated tumour cells from the bone marrow of early stage breast cancer patients – Results from a large single-centre analysis. Eur J Cancer 2014; 50: 2550-2559
• 37 Cohen EN, Jayachandran G, Hardy MR. et al. Antigen-agnostic microfluidics-based circulating tumor cell enrichment and downstream molecular characterization. PLoS One 2020; 15: e0241123
• 38 Weidele K, Stojanović N, Feliciello G. et al. Microfluidic enrichment, isolation and characterization of disseminated melanoma cells from lymph node samples. Int J Cancer 2019; 145: 232-241
• 39 Reinhardt K, Stückrath K, Hartung C. et al. PIK3CA-mutations in breast cancer. Breast Cancer Res Treat 2022; 196: 483-493
• 40 Turner NC, Oliveira M, Howell SJ. et al. Capivasertib in Hormone Receptor-Positive Advanced Breast Cancer. N Engl J Med 2023; 388: 2058-2070
• 41 Lambert A, Salleron J, Lion M. et al. Comparison of Three Real-Time PCR Assays for the Detection of PIK3CA Somatic Mutations in Formalin-Fixed Paraffin Embedded Tissues of Patients with Breast Carcinomas. Pathol Oncol Res 2019; 25: 1117-1123
• 42 Thulin A, Andersson C, Werner Rönnerman E. et al. Discordance of PIK3CA and TP53 mutations between breast cancer brain metastases and matched primary tumors. Sci Rep 2021; 11: 23548
• 43 Deng G, Krishnakumar S, Powell AA. et al. Single cell mutational analysis of PIK3CA in circulating tumor cells and metastases in breast cancer reveals heterogeneity, discordance, and mutation persistence in cultured disseminated tumor cells from bone marrow. BMC Cancer 2014; 14: 456
• 44 Park J, Cho SY, Chang ES. et al. Analysis of PIK3CA Mutation Concordance and Frequency in Primary and Different Distant Metastatic Sites in Breast Cancer. Cancer Res Treat 2023; 55: 145-154
• 45 Wang Y, Li X, Zhang S. et al. Analysis of PIK3CA mutations in the primary and recurrent tumors of hormone receptor positive/human epidermal growth factor receptor 2 negative breast cancer. Jpn J Clin Oncol 2024; 54: 1024-1031
• 46 Phan TG, Croucher PI. The dormant cancer cell life cycle. Nat Rev Cancer 2020; 20: 398-411

Publication History

Received: 27 August 2025

Accepted after revision: 22 November 2025

Article published online:
13 January 2026