The Correlation between Plasma Circulating Tumor DNA and Radiographic Tumor Burden

Abstract

Conventional blood-based biomarkers and radiographic imaging are excellent for use in monitoring different aspects of malignant disease, but given their specific shortcomings, their integration with other, complementary markers such as plasma circulating tumor DNA (ctDNA) will be beneficial toward a precision medicine–driven future. Plasma ctDNA analysis utilizes the measurement of cancer-specific molecular alterations in a variety of bodily fluids released by dying tumor cells to monitor and profile response to therapy, and is being employed in several clinical scenarios. Plasma concentrations of ctDNA have been reported to correlate with tumor burden. However, the strength of this association is generally poor and highly variable, confounding the interpretation of longitudinal plasma ctDNA measurements in conjunction with routine radiographic assessments. Herein is discussed what is currently understood with respect to the fundamental characteristics of tumor growth that dictate plasma ctDNA concentrations, with a perspective on its interpretation in conjunction with radiographically determined tumor burden assessments.

Key Points.

• •Cellular turnover rate likely plays a significant role in circulating tumor DNA (ctDNA) production through cellular death rate.
• •The value of circulating tumor DNA as a biomarker for disease surveillance relies upon its relationship to proliferation.
• •Longitudinal monitoring of ctDNA in cancer patients can provide valuable insight into tumor cell turnover and proliferation through direct measurement of changes in cell death rate.

Circulating tumor DNA (ctDNA) is a valuable clinical tool to support cancer care. Presently, plasma ctDNA analysis is being employed in several clinical scenarios, including monitoring for residual minimal disease in colorectal cancer,¹ monitoring for the emergence of resistance mutations in patients with lung cancer,² and serving as a surrogate DNA source for mutation profiling as an alternative to tumor tissue biopsy³ (ie, liquid biopsy). Kinetic changes in plasma ctDNA concentration ([ctDNA]) within the first weeks to months of therapeutic monitoring have demonstrated prognostic utility and are increasingly being applied to patient care.4, 5, 6, 7 In contrast, kinetic changes in plasma [ctDNA] in the later stages of therapeutic monitoring (once a patient is established on treatment) to detect progression is complicated by a lack of understanding about the relationship between plasma [ctDNA] and radiographically determined tumor burden measurements, which remain the standard of care for determining late-stage disease progression in solid tumors. Additionally, the use and interpretation of ctDNA measurements are further confounded given that a patient's radiologic assessments and blood draws for ctDNA detection, in practice and in the published literature, may not necessarily take place during the same visit. Herein is discussed what is currently understood with respect to the fundamental characteristics of tumor growth that dictate plasma [ctDNA], with a perspective on its interpretation in conjunction with radiographically determined tumor burden assessments. These reflections led to the conclusion that plasma [ctDNA] are primarily driven by a proliferating tumor burden. The implication of this conclusion supports the premise that sustained and increasing plasma [ctDNA] reflect clinical progression and suboptimal therapy prior to radiographically evident increases in tumor mass, and supports the application of longitudinal plasma [ctDNA] to aid in the early detection of disease progression and recurrence.^8,9

More than 60% of patients diagnosed with nonmetastatic disease are at risk for eventual disease recurrence.¹⁰ The risk for recurrence depends on cancer type and can range from approximately 17% in colorectal cancer, to >75% in glioblastoma.11, 12, 13 Furthermore, in >30% of patients with advanced melanoma, the disease ultimately progresses despite first-line therapy, and 52% of patients with stage III melanoma may experience recurrence despite undergoing adjuvant therapy.14, 15, 16, 17 Although certain risk factors can be used to identify patients whose disease is more likely to recur or progress, many patients who present with recurrence are asymptomatic.^18,19 Currently, radiographic imaging is the clinical tool most universally used to assess solid tumor progression.²⁰ Response Evaluation Criteria in Solid Tumors (RECIST; version 1.1) guidelines have been established to standardize the criteria for determining whether a given tumor is responding, stable, or progressing, depending on a change in size as determined by cross-sectional imaging. RECIST criteria are applied to radiographic measurements by computed tomography or positron emission tomography to determine whether the patient shows complete response (<10 mm all lesions), partial response (30% decrease in sum diameter), progressive disease (20% increase in sum diameter), or stable disease (insignificant size change).²⁰ Positron emission tomography–based radiography increases sensitivity by using an ¹⁸F-fluorodeoxyglucose substrate to evaluate metabolically active tumor cells.^21,22 However, radiographic surveillance has limitations, including low frequency of monitoring, inability to monitor longitudinal molecular heterogeneity, patient exposure to ionizing radiation, and a low sensitivity in recurrence detection.23, 24, 25

Cell-free DNA (cfDNA), and its tumor constituent, ctDNA, has gained traction in the literature on clinical diagnostics and as a tool in clinical trials over the past decade.26, 27, 28, 29, 30, 31, 32 This liquid-biopsy approach utilizes the measurement of cancer-specific molecular alterations in a variety of bodily fluids released by dying tumor cells to monitor and profile the response to therapy. Research has shown that ctDNA quantities, expressed either as a percentage of variant-allele frequency or as the concentration of copies per milliliter of body fluid, are associated with prognosis and overall survival.33, 34, 35, 36, 37 However, ctDNA quantities have also shown inconsistencies in relation to other blood-based biomarkers and radiographic imaging.38, 39, 40, 41 Currently, a lack of understanding about the basis of any correlation between ctDNA and other well-established diagnostic techniques may be stifling its incorporation into widespread clinical use. It is argued that the underlying factor dictating the relationship between plasma ctDNA and tumor burden is tumor proliferation.

ctDNA Origins and Kinetics

cfDNA is a collective term used to describe small DNA fragments that circulate freely in bodily fluids, and may be of nuclear or mitochondrial origin.³¹ These fragments originate largely from necrosis, apoptosis, and cellular turnover of hematopoietic cells.42, 43, 44, 45, 46 As cancer cells die, cfDNA containing tumor-specific genetic alterations can enter the systemic circulation and is detected as ctDNA.^31,32,47 ctDNA is observed either in its free form, associated with single histones, within larger nucleosome complexes, or within microparticles such as exosomes and apoptotic bodies.^31,48 ctDNA is rapidly eliminated in vivo, with a median half-life (t_1/2) of 35 minutes.⁴⁹

The fact that apoptosis and or necrosis of tumor tissue is the source of plasma ctDNA, and that plasma ctDNA is rapidly eliminated, would lead to the expectation that patients with elevated concentrations of plasma ctDNA would demonstrate reductions in tumor mass. However, in practice, increasing tumor burden is typically associated with increasing plasma [ctDNA],⁵⁰ but the correlation between tumor burden and ctDNA is poor to moderate, with correlation coefficients on the order of (r² ≈ 0.5). This poor degree of correlation limits the utility of plasma ctDNA measurements used for predicting tumor burden and confounds the interpretation of plasma ctDNA measurements. In recent work it was observed that the ratio of plasma [ctDNA] to tumor burden was approximately 5-fold greater, with a significant correlation with tumor burden (r² = 0.91) under conditions of worsening disease.⁵¹ These data led to the exploration of the underlying factors that influence the relationship between plasma ctDNA and tumor burden.

Unless otherwise specified by the study design, most plasma ctDNA measurements published in the literature, and in particular those measurements obtained during maintenance systemic therapy, represent what can be considered a random steady-state measurement. Pharmacokinetic principles dictate that in order to maintain steady-state plasma concentrations of a substance with rapid elimination, a corresponding rate of infusion is required.

The following example is provided to illustrate the conditions necessary for sustaining elevated plasma concentrations of ctDNA. Assuming a volume of distribution (V_d) of 5 L (blood compartment) and a median t_1/2 of ctDNA (0.58 hours) from lung cancer patients after tumor resection observed by Chen,⁴⁹ the clearance (CL) of ctDNA can be approximated, using Equation 1, as 5.96 L/hour.

$$\mathrm{CL}=\left(0.692\ast V_{\mathrm{d}}\right)/{\mathrm{t}}_{1/2}$$ (1)

For ctDNA to maintain steady-state concentration (C_ss), a compensatory infusion rate from the tumor must also be present (Equation 2).

$$C_{\mathrm{ss}}=\text{Infusion}\text{rate}/\text{Elimination}\text{rate}$$ (2)

Because tumor cell death is likely the primary origin of ctDNA,^28,30,52 the infusion rate must be proportional to the tumor cell death rate (Equation 3). (Note: Here, it is assumed that the amount of ctDNA released by living cells and circulating tumor cells is negligible). Thus, if ctDNA is measurable, then there must be a proportional amount of continual cell death occurring in the tumor:

$$C_{\mathrm{ss}}=x\left(\text{Death}\text{rate}\right)/\text{Elimination}\text{rate}$$ (3)

where x represents the combined effects of other factors, such as ctDNA released by living tumor cells and circulating tumor cells, DNA shedding probability as described by Avanzini et al,⁵³ and ctDNA that does not escape the tumor microenvironment due to the phagocytosis of apoptotic bodies by neighboring cells and macrophages.⁵⁴ These factors are still not fully understood; therefore, x = 1 is assumed for simplicity.

The range of plasma ctDNA (copies/mL) observed by Bettegowda et al⁵⁵ among various cancer types was 1 to 200,000 copies/mL. Using Equation 3, sustaining a steady-state concentration of 1 copy/mL of ctDNA would necessitate a cell death rate of 24,000 cells/day. Late-stage colorectal cancer in which plasma [ctDNA] may exceed 20,000 copies/mL would suggest a cell death rate of 4.8 × 10⁸ cells/day. A tissue volume of 1 cm³ is approximately 1 × 10⁸ to 1 × 10⁹ cells.⁵⁶ This indicates that the cell death rate would result in a decrease in tumor size of 0.48 to 4.8 cm³/day. Paradoxically, patients with late-stage cancer, metastasis, and elevated plasma [ctDNA] frequently exhibit increasing tumor burden on radiography.

Life and Death

Cellular turnover is a crucial component of homeostasis in healthy tissue. The lifespan of human eukaryotic cells can vary from 5 days in the case of colon epithelial cells,⁵⁷ to 100 days in erythrocytes,⁵⁸ to 200 to 300 days in liver cells,⁵⁹ and persisting indefinitely in the case of neuronal cells.⁶⁰ Hematopoietic cells represent 85% of total cellular turnover per day⁵⁷ while contributing nearly 85% of total cfDNA in healthy individuals.⁴⁶ Furthermore, the cell types associated with cancers demonstrating the highest concentrations of ctDNA (gastrointestinal, epithelial) possess some of the highest proliferation rates in the body outside of hematopoietic cells.^55,57

Considering this observation, the cellular turnover rate likely plays a significant role in cfDNA/ctDNA production through cell death rate, as described in Equation 4:

$$\text{Cellular}\text{turnover}\text{rate}\propto \text{Proliferation}\text{rate}-\text{Death}\text{rate}$$ (4)

Although ctDNA may reflect cell death and, indirectly, cellular turnover, its value as a biomarker in disease surveillance relies upon its relationship to proliferation. Dysregulation between proliferation and apoptosis is the driving force of tumorigenesis.⁶¹ However, complete ablation of apoptosis with uncontrolled proliferation would lead to a tumor volume of 1 cm³ (approximately 1 × 10⁹ cells), increasing to 1000 cm³ (a tumor incompatible with life) within 10 cell-doubling events.⁶² Thus, continual cell death must play a role in an actively proliferating tumor in order to maintain sustainable growth.

Tumor cell death via apoptosis and necrosis have been identified as the most prominent sources of ctDNA.^28,30,52 In 2001, Liu et al⁶³ quantified proliferation and apoptosis in breast cancer biopsy samples and found that increased apoptotic cell counts were strongly associated with tumor grade, size, metastasis, and shortened disease-specific survival. Similar observations have been reported in colorectal cancer,⁶⁴ endometrial adenocarcinoma,⁶⁵ and prostate cancer.⁶⁶ Apoptosis-induced proliferation is an established link wherein apoptotic cells produce and secrete a burst of mitogens, signaling proliferation in neighboring cells.67, 68, 69 Passive cell death through necrosis may play a smaller, but important, role in proliferation signaling via the induction and enhancement of apoptosis.70, 71, 72 The understanding of the mechanisms and pathways of cell death has increased greatly in the past two decades.⁷³ An intermediary between apoptosis and necrosis—necroptosis, wherein cells undergoing apoptosis, remain undead, and continually secrete mitogens—may play a significant role in tumor proliferation and metastasis.^68,74,75 The significance and influence that lesser established death mechanisms (pyroptosis, ferroptosis, autosis, parthanatos) have on tumorigenesis and ctDNA release kinetics are currently unknown.

ctDNA, Tumor Burden, and Disease Progression

Using a simple generalization, an increase in the proliferation rate of malignant cells enables disease progression. To facilitate the earliest possible detection of progression or recurrence, then, requires the earliest possible detection of a change in proliferation rate. Highly sensitive, direct measurement of in vivo tumor proliferation rate is currently not achievable. Additionally, radiographic tumor burden size is not a direct measure of proliferation rate.

Figure 1 is a schematic representation of how plasma concentrations of ctDNA are influenced by the various stages of tumor development, therapeutic response, and recurrence.

Figure 1.

A model of tumor progression and ctDNA release. Panel 1: At the initiation stage of tumor proliferation, growth rate predominates, with minimal cell death, resulting in ctDNA concentrations ([ctDNA]) that are less than the lower limit of detection (LOD). Tumor proliferation leads to the formation of a nutrient-deprived necrotic core (gray cells). Panel 2: Cancer cells proliferate and increase in abundance. Cell death rate increases and low [ctDNA] may be detectable at this stage (brown region). Panel 3: At later stages of disease, the tumor microenvironment is fully established, and cell death rate increases dramatically. Increased cellular turnover allows the opportunity for the malignancy to differentiate and become heterogenous as invasive cancer cells appear (darkred regions). Panel 4: The initiation of therapy promotes tumor cell death and diminishes cellular proliferation, leading to rapid clearance of ctDNA due to the absence of proliferation. Panel 5: Residual proliferation after therapy allows for tumor subclones to reestablish themselves, including new therapy-resistant cells. Panel 6: New therapy-resistant subclones begin yielding ctDNA as proliferation increases (green region).

Tumor burden is maintained by the balance between proliferation and cell death (Equation 5).

$$\Delta \text{TB}\propto \Delta r_{\mathrm{g}}-\Delta r_{\mathrm{d}}$$ (5)

where TB is the tumor burden, r_g is the cell proliferation rate, and r_d is the cell death rate.

This relationship is illustrated in Figure 2, where relative tumor proliferation rates can be inferred at various stages after treatment, and the cell death rate is represented by [ctDNA]. In a scenario of increasing tumor burden, the proliferation rate exceeds the cell death rate. A clinical scenario in which the proliferation rate is increasing with a proportionate increase in cell death rate would result in no change in tumor mass, and radiographic tumor burden would fail to detect progression. This scenario would appear as stable disease in RECIST, despite increasing proliferation.

Figure 2.

Two theoretical scenarios—disease recurrence and complete therapeutic response—comparing the changes in the rates of tumor proliferation and cell death (represented by ctDNA concentration [ctDNA]), both in arbitrary units. Given the relationship ΔTB ∝ Δr_g – Δr_d ([ctDNA]), where TB is the tumor burden, r_g is the proliferation rate, and r_d is the cell death rate, the relative tumor burden can be inferred at various stages after treatment and during disease surveillance, based on the relative rates of tumor proliferation and [ctDNA] release. Arrows indicate the time points relevant in the calculation of Δ. The theoretical lower limit of detection (LOD) of ctDNA is represented by the red line; theoretical radiographic LOD is represented by the blue line. A: Partial response with recurrence. Prior to treatment, proliferation is increased (Δr_g = 20), resulting in an increased rate of cell death (Δr_d = 5). The net effect is an increase in ctDNA, with a net increase in TB of 15 after 1 month. (A1): At 1 month after treatment, proliferation is partially reduced (Δr_g = −40), resulting in a reduction in cell death rate (Δr_d = −30), driving decreases in both TB (ΔTB = −10) and plasma [ctDNA]. (A2): Several months after treatment, proliferation remains reduced (Δr_g = 0) and the cell death rate normalizes to the proliferation rate (Δr_d = 0), resulting in persistently low but detectable quantities of plasma ctDNA without any change in TB (ΔTB = 0) relative to the previous month. (A3): Developing resistance to therapy allows for an increased rate of proliferation (Δr_g = 20), which drives an increase in cell death rate (Δr_d = 15), resulting in an acute increase in plasma ctDNA and a gradual increase in TB (ΔTB = 5) relative to the previous month. B: Complete response. As in A, prior to treatment, proliferation increases (Δr_g = 20), resulting in an increased rate of cell death (Δr_d = 5). The net effect is an increase in ctDNA, with an increase in TB of 15 after 1 month. (B1): At 1 month after treatment, proliferation is dramatically reduced (Δr_g = −110), resulting in a reduction in cell death rate (Δr_d = −70), such that plasma [ctDNA] are below the LOD, and TB is also reduced (ΔTB = −40). (B2): Several months after treatment, the rates of proliferation and cell death remain unchanged, resulting in persistently undetectable plasma [ctDNA] and no change in TB relative to the previous month. (B3): Complete response ablates proliferation (Δr_g = −20), and death rate persists (Δr_d = 0), resulting in continued concentrations of plasma ctDNA below the LOD and a reduction in residual TB (ΔTB = −20) relative to the previous month.

In the scenario of the misidentification of an increased proliferation rate as stable disease, ctDNA would help to elucidate the previously undetectable change in cellular turnover and proliferation due to increasing [ctDNA] caused by an increasing cell death rate. This was observed clinically in patients with colorectal cancer, in whom increases in plasma [ctDNA] preceded radiographically identified progressive disease by up to 11.5 months.76, 77, 78 In patients with esophageal adenocarcinoma and non–small-cell lung cancer, increases in plasma ctDNA preceded radiographically identified progressive disease by 100 days and a median of 88 days, respectively.^79,80 An additional beneficial aspect of ctDNA is in predicting overall disease-free survival after treatment based on plasma ctDNA clearance rate.⁸¹ ctDNA clearance is the net measure of infusion and elimination. Immediately after the initiation of therapy, [ctDNA] can show varied clearance kinetics based on therapy.⁸¹ For instance, after surgical intervention, the removal of bulk cells can lead to deceases in [ctDNA] of up to 99%.^82,83 This decrease most likely reflects rapid elimination in the absence of a tumor tissue source. The initiation of radiation therapy or chemotherapy may lead to transient peaks in [ctDNA] in some patients within the first 72 hours, followed by a sharp decrease.84, 85, 86, 87 This scenario most likely reflects a rapid and transient infusion from dying cells, followed by the absence of continued infusion from the responding tumor. Patients can have varied responses to immunotherapy, with rapid clearance, fluctuations/rises within 3 to 16 weeks, or initial rapid clearance followed by a steady increase in concentration. Each of these scenarios would be representative of the net effect of therapy on the proliferation state of the tumor. A reduced plasma ctDNA clearance rate after treatment, and failure to completely ablate measurable ctDNA, are also indicative of poor progression-free survival.⁸¹ This is consistent with a scenario in which tumor proliferation capacity has not been fully eliminated. A complete response to therapy leads to sharp and complete decreases in ctDNA,^49,83 which is consistent with the absence of tumor proliferation and rapid elimination of the released ctDNA. A partial response to therapy leads to incomplete or slow clearance of ctDNA, most likely the result of a partial reduction in tumor-proliferation rate, with no change in elimination. As established earlier, ctDNA is a component of cellular turnover and related to proliferation. Thus, residual ctDNA indicates residual tumor cellular turnover and proliferation after treatment, while the complete ablation of ctDNA indicates ablation of cellular turnover and proliferation.

Initially, one may associate decreasing tumor burden with increasing cell death rate and expect increased [ctDNA]; however, decreasing tumor burden indicates only that the cell death rate exceeds the proliferation rate; it cannot be used as a surrogate for an absolute value of either rate. Due to the rapid elimination (short t_1/2) of ctDNA, complete response and decreasing tumor burden quickly deplete stores of DNA available for shedding into the circulation, the result being great reductions in both rates, with the cell death rate exceeding the proliferation rate, allowing for a decrease in tumor burden. If the ctDNA stores are not depleted after therapy, there must be a replenishing mechanism via residual cellular turnover and proliferation. This residual tumor activity lays the foundation for disease recurrence. Cell death and proliferation rates exist in an equilibrium that drives tumor dynamics and cellular turnover. Two theoretical patient scenarios, complete response and partial response (Figure 2), are further elaborated in the context of cellular turnover (Figure 3). A complete response to therapy yields a large reduction in tumor cellular turnover and longer disease-free survival (Figure 3), while a partial response to therapy maintains residual proliferation activity despite an initial reduction in tumor burden (Figure 3).

Figure 3.

Graphical representation of the relationship between the rates of cell proliferation and death in the context of changes in tumor burden and tumor cell turnover. A: The diagonal line represents stable tumor burden as the rates of cell proliferation and death increase proportionately with each other. In the blue region, the cell death rate is high, while the proliferation rate is low, which reduces tumor burden. In the red region, the proliferation rate is high, while the cell death rate is low, which increases tumor burden. The shaded arcs represent zones of cellular turnover that are low (bottom left) and high (top right). The numbered points represent longitudinal radiologic measurements of disease status. B: 1 → 2, Disease burden remains stable despite an increased proliferation rate (Δr_g = Δr_d [ctDNA]; increasing in parallel); 2 → 3, the proliferation rate begins to surpass the cell death rate, resulting in increased radiographic tumor burden (Δr_g > Δr_d); 3 → 4, a complete response to therapy greatly reduces the rate of tumor cell turnover, resulting in a steep decrease in the rates of cell proliferation and death [Δr_g < Δr_d; Δr_d < lower limit of detection (LOD)]. C: 1 → 3, Clinical progression identical to that in the previous case; 3 → 4, a partial response to therapy leads to residual proliferation and continued cellular turnover (Δr_g < Δr_d; Δr_d > LOD); 4 → 5, continued cellular turnover generates subpopulations of highly proliferative cells, leading to tumor recurrence (Δr_g > Δr_d). Theoretical ctDNA LOD is represented by the red line; theoretical radiographic LOD is represented by the blue line.

Conclusion

Due to its rapid clearance, ctDNA represents a continuously updating snapshot of the rate of tumor cell death. Longitudinal monitoring of ctDNA in cancer patients can provide valuable insight into tumor cell turnover (activity) and proliferation through direct measurement of changes in the cell death rate. Theoretically, stable tumor burden with detectable ctDNA suggests proliferation in excess of death and an increased risk for recurrence; increasing ctDNA with stable tumor burden indicates a parallel increase in death rate and proliferation rate, signifying imminent recurrence; and stable tumor burden with no detectable ctDNA indicates minimal cellular turnover and a low risk for recurrence. Conventional blood-based biomarkers and radiographic imaging are excellent for monitoring the aspects of disease, but each has specific shortcomings that prevent full integration into a precision medicine–driven future. Despite inconsistent correlation with various cancer markers and variations in levels detected among different tumor types, changes in ctDNA quantities frequently precede changes in radiographically determined tumor burden and methods of cancer biomarker surveillance when used to predict or detect disease recurrence. This facet may be attributed to ctDNA's unique representation of the balance between proliferation and cell death, which can result in increased sensitivity in detection compared to radiographic imaging. However, as with other blood-based biomarkers, ctDNA presents challenges regarding reproducibility and harmonization, which may limit widespread adoption by clinics.88, 89, 90, 91 When tumors are small, the signal-to-noise ratio is low due to the presence of a high background made of wild-type circulating DNA, which can limit the analytical sensitivity of the assay.⁸⁹ Biological factors, such as differences in the degree and rate of association or disassociation with various intracellular and extracellular components, may cause ctDNA to be cleared from the circulation at rates that differ between patients.⁹¹ Additionally, pre-analytical discrepancies in biospecimen collection, processing, and storage practices are an ongoing challenge that prohibits widespread adoption by clinics.⁹⁰ Even so, ctDNA provides substantial technical value to augment radiographic surveillance techniques. The theoretical lower limit of detection of ctDNA, which is limited by cell death events and pharmacokinetic parameters, is multiple orders of magnitude lower than tumor burden, which is currently confined to millimeter-based measurements. Temporal resolution of surveillance is also improved with the use of a blood-based biomarker, which may lessen the time required for detecting recurrence. In the coming decade, ctDNA is poised to become a widely used, powerful surveillance tool that aides in the understanding and early detection of tumor recurrence and therapeutic resistance.

Disclosure Statement

None declared.