Review Reports - Elevated D-Dimer Levels in Older Medical Emergency Department Patients: Real-Life Data on Associations with Severe Acute Medical Problems and Occult Malignancy

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Elevated D-dimer levels in older medical emergency department patients. Real-life data on association with severe acute medical problems and occult malignancy

Summary: The aim of this study was to investigate the diseases associated with elevated D-dimer levels, with a special focus on occult malignancy in older medical ED patient.

Critique: This is an interesting study investigating the commonly ordered D-dimer test in the Emergency Department. The study has some merit for publication; however, the methods and results need clarification.

D-dimer is primarily ordered in the Emergency Department (ED) to rule out a Pulmonary Embolus (PE), however, this study had a special focus on occult malignancy, so it is unclear to me why the ED as chosen to collect the samples. Regarding the data collection, the authors state “We collected the following diagnoses present during the ED visit, which are known to be associated with elevated D-dimer levels: sepsis, infection, inflammation and ischemia.” It seems to me that all elderly patients presenting to the ED regardless of chief complaint, should have D-dimer collected. The described methods seem to collect D-dimers on patients that are already known to have elevated D-dimers, which would increase the likelihood to diagnose occult malignancies.

Secondly, the style of writing is confusing; “Data were collect from patients’ medical records……..” I am assuming the authors mean clinical data and the D-dimers were collected at time of presentation to the ED. Patients were enrolled from July 2016 until February 2017, over eight years ago, why the delay in attempting publication?

Another confusing statement, “…were assessed and treated by an internist or gastroenterologist……….” I thought this study was performed in an ED, so were Emergency Physicians involved?

Results: The authors state that 31 patients were diagnosed with a malignancy (occult?) within 6 months of follow up. A variety of malignancies were seen with colorectal and prostate carcinoma. Were most of these occult malignancies diagnosed with standard methods, colposcopy, EGD, PSA, CT scan? If so, what triggered the decision to do these procedures or tests? I agree with the authors” conclusion that D-dimers should not be used as a screening method for carcinoma.

Discussion and Conclusions: Sections are well-written

Author Response

We sincerely thank the reviewers for the constructive feedback. Your comments improved the overall quality of the manuscript.

Comment1: D-dimer is primarily ordered in the Emergency Department (ED) to rule out a Pulmonary Embolus (PE), however, this study had a special focus on occult malignancy, so it is unclear to me why the ED as chosen to collect the samples. Regarding the data collection, the authors state “We collected the following diagnoses present during the ED visit, which are known to be associated with elevated D-dimer levels: sepsis, infection, inflammation and ischemia.” It seems to me that all elderly patients presenting to the ED regardless of chief complaint, should have D-dimer collected. The described methods seem to collect D-dimers on patients that are already known to have elevated D-dimers, which would increase the likelihood to diagnose occult malignancies.

Response 1: We thank the reviewer for this comment. To clarify, D-dimer levels were measured in all older ED patients who participated in the study, regardless of their presenting complaint or the presence of sepsis, infection, inflammation or ischemia. The conditions listed in the Methods section were collected as additional variables known to be associated with elevated D-dimers, but they were not criteria for inclusion nor were D-dimers measured selectively in these patients. We adjusted the Data Collection section (Page 2, line 81/83 and Page 3, line 94-96)so this is clear.

Comment 2: Secondly, the style of writing is confusing; “Data were collect from patients’ medical records……..” I am assuming the authors mean clinical data and the D-dimers were collected at time of presentation to the ED.

Response 2: We adjusted the data collection section so this is more clear (Page 2, line 81/83).

Comment 3: Patients were enrolled from July 2016 until February 2017, over eight years ago, why the delay in attempting publication?

Response 3: We appreciate the reviewer’s comment. The present analysis was part of a larger research project investigating predictors of adverse outcomes (including mortality) in older ED patients, which included several related substudies. The analysis focusing on occult malignancy was conducted later within this framework, which explains the delay in submission. Nevertheless, we believe the findings remain clinically relevant, as D-dimer continues to be widely used in the diagnostic work-up of older ED patients.

Comment 4: Another confusing statement, “…were assessed and treated by an internist or gastroenterologist……….” I thought this study was performed in an ED, so were Emergency Physicians involved?

Response 4: In the Netherlands, patients presenting to the ED with internal medicine or gastroenterology–related complaints are assessed by internists (internal medicine physicians), not by emergency physicians. Most of these patients are referred to the ED by their general practitioner. The internists have been trained in emergency medicine and work in a similar way as emergency physicians.

Comment 5: Results: The authors state that 31 patients were diagnosed with a malignancy (occult?) within 6 months of follow up. A variety of malignancies were seen with colorectal and prostate carcinoma. Were most of these occult malignancies diagnosed with standard methods, colposcopy, EGD, PSA, CT scan? If so, what triggered the decision to do these procedures or tests? I agree with the authors” conclusion that D-dimers should not be used as a screening method for carcinoma.

Response 5: We thank the reviewer for this helpful comment. Most malignancies were diagnosed based on clinical symptoms, as D-dimer values were blinded to the treating physicians. Diagnostic procedures were performed based on clinical findings and included CT scans, endoscopy (EGD/colonoscopy), PET scans, PSA testing, and other standard investigations as clinically indicated. There was no standardized work-up in our cohort aimed at detecting or excluding malignancy. Instead, when deemed appropriate—based on clinical findings and with patient consent—ancillary investigations were performed. It is important to note that D-dimer levels were not available to the treating physicians at the time of these clinical decisions. Therefore, the detection of malignancy was entirely guided by symptoms and clinical judgment.

Comment 6: Discussion and Conclusions: Sections are well-written

Response 6: We thank the reviewer for the positive feedback and are glad that the Discussion and Conclusions were found clear and well-written.

Reviewer 2 Report

Comments and Suggestions for Authors

Manuscript: ecm-3811018

Critique

Acquired thrombophilias develop later in life, and include, for instance, antiphospholipid syndrome as well as several other disorders and conditions that impact coagulation/fibrinolytic systems. The paradigmatic clinical expression of acquired thrombophilia is an unprovoked deep venous thrombosis (DVT), routinely diagnosed by ultrasound. Elevated D-dimer is typically used to confirm the diagnosis. It must be understood that notwithstanding independent activation of fibrinolysis, D-dimer is a marker of an increase in coagulation due to a wide spectrum of diseases.

Based on previous studies (briefly reviewed in the Introduction), the authors postulate elevated D-dimer levels in older medical ED patients, irrespective of the presenting clinical problem, may be associated with occult malignancy.

In this study, D-dimer levels are obtained in subjects admitted to the ED with a median age of 79 years (life expectancy in the Netherlands in 2016 is reported to have been 81.6 yrs), and a range of acute and chronic disease processes (for example, sepsis, infection, and active malignancy). The authors find that in older ED patients, elevated D-dimers are not specific to VTE and that other conditions including malignancy, ischemia and sepsis need to be considered. Although interesting, the authors’ findings are neither novel, nor particularly compelling.

A more direct approach to examination of the authors’ hypothesis would be to screen older patients for elevated D-dimer who are not in the ED for any acute, generally systemic process. Suggest the authors address this limitation of their study.

Technically the authors should refer to patients as subjects once enrolled in the study.

Specific Questions|Suggestions

Page 2, line 61: Suggest deleting, real-life. RE: Phrase lacks specific meaning.

Page 3, line 123: Analysis includes 407 patients enrolled at a single center during a 7-month period in the latter half of 2016 and first two months of 2017. Arguably current diagnostic techniques and sensitivities have substantially improved during the ensuing 7 years. However, analyses of subjects are exemplary.

Page 6, lines 195-196 and Table 3: Of subjects without a diagnosis of active malignancy at the time of their ED visit, 31 (9.2%) were diagnosed with a malignancy within 6 months of follow up. Of these, 5 had colorectal carcinoma and 4 were found to have prostatic carcinoma. It would be interesting to know if any of these had undergone screening colonoscopy or had measurement of PSA within 6 months prior to ED admission. Are these data obtainable?

Page 9, lines 313-323: The authors conclude that D-dimer as a screening tool for occult malignancy lacks sufficient sensitivity. Would the authors further speculate that D-dimer possibly serves as a tumor marker, whereby it reflects differences in certain clinicopathologic features and/or prognosis as proposed for say breast carcinoma [PLoS One. 2019;14(9):e0221374].

Author Response

Comment 1: In this study, D-dimer levels are obtained in subjects admitted to the ED with a median age of 79 years (life expectancy in the Netherlands in 2016 is reported to have been 81.6 yrs), and a range of acute and chronic disease processes (for example, sepsis, infection, and active malignancy). The authors find that in older ED patients, elevated D-dimers are not specific to VTE and that other conditions including malignancy, ischemia and sepsis need to be considered. Although interesting, the authors’ findings are neither novel, nor particularly compelling.

Response 1: We thank the reviewer for this comment and acknowledge that our study is limited to older patients presenting to the ED. In the ED, elevated D-dimers frequently prompt discussions regarding further work-up, including exclusion of VTE or consideration of occult malignancy. In this context, we believe our study adds valuable information, as D-dimers were measured in all participants regardless of their chief complaint or suspicion of VTE, providing insight into the association between D-dimer and occult malignancy in a real-world clinical setting.

We added following to the limitation section (Page 11, Line 346-350): “In addition, the study was conducted in older patients presenting to the ED with a range of acute and chronic conditions, which limits generalizability to healthier populations. Nevertheless, D-dimers were measured in all participants regardless of chief complaint or suspicion of VTE, allowing assessment of the association between D-dimer and occult malignancy in a real-world clinical ED setting.”

Comment 2: Technically the authors should refer to patients as subjects once enrolled in the study.

Response 2: We thank the reviewer for this comment. In our manuscript, we used the term “patients” for clarity and readability. Not all participants underwent imaging; however, malignancy diagnoses were captured either through clinical assessment or within six months of follow-up. We believe this terminology accurately reflects the clinical context and intention of the study.

Specific Questions|Suggestions

Comment 3: Page 2, line 61: Suggest deleting, real-life. RE: Phrase lacks specific meaning.

Response 3: Thank you for this suggestion, we deleted this phrase from the introduction section.

Comment 4: Page 3, line 123: Analysis includes 407 patients enrolled at a single center during a 7-month period in the latter half of 2016 and first two months of 2017. Arguably current diagnostic techniques and sensitivities have substantially improved during the ensuing 7 years. However, analyses of subjects are exemplary.

Response 4: We appreciate the reviewer’s comment and agree that diagnostic techniques have evolved since 2016–2017. Nevertheless, this study provides valuable insights into the association of D-dimer with occult malignancy. In our cohort, most malignancies were diagnosed based on imaging techniques (e.g., CT scans, PET scans), endoscopies, and pathology reports. Given that most cancers manifest with symptoms within six months, we believe that the majority of occult malignancies were likely identified, despite the study period being several years ago.

Comment 5: Page 6, lines 195-196 and Table 3: Of subjects without a diagnosis of active malignancy at the time of their ED visit, 31 (9.2%) were diagnosed with a malignancy within 6 months of follow up. Of these, 5 had colorectal carcinoma and 4 were found to have prostatic carcinoma. It would be interesting to know if any of these had undergone screening colonoscopy or had measurement of PSA within 6 months prior to ED admission. Are these data obtainable?

Response 5: Thank you for this fine suggestion. Unfortunately, this is not obtainable for us since data regarding population screening tests are not automatically incorporated into the patients’ medical records. In addition, we do not have data regarding PSA measurements prior to the ED visit.

Comment 6: Page 9, lines 313-323: The authors conclude that D-dimer as a screening tool for occult malignancy lacks sufficient sensitivity. Would the authors further speculate that D-dimer possibly serves as a tumor marker, whereby it reflects differences in certain clinicopathologic features and/or prognosis as proposed for say breast carcinoma [PLoS One. 2019;14(9):e0221374].

Response 6: We thank the reviewer for this insightful suggestion and corresponding reference. Our findings, that D-dimer is associated with malignancy, supports the findings in the meta-analysis that found that D-dimer is associated with advanced disease and metastasis in breast cancer patients. However, since D-dimer can also be elevated by other conditions such as infection or VTE, which are common in patients with advanced cancer, it likely lacks specificity and is therefore unlikely to serve as an ideal tumor marker for cancer screening or follow-up. Nevertheless, we agree that in patients with markedly elevated D-dimer levels, the possibility of an underlying malignancy should be considered.

We adjusted the discussion and future perspective sections as follows (Page 11, Line 246-350): “One meta-analysis found that elevated D-dimer levels are associated with advanced disease and metastasis in breast cancer patients suggesting that it could be a potential tumor marker[23]. However, due to lack of specificity D-dimers are unlikely to serve as an ideal tumor marker for cancer screening or follow-up. Nevertheless, in patients with markedly elevated D-dimer levels, the possibility of an underlying malignancy should be considered.”.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript addresses an important and very pragmatic question: how to interpret elevated D-dimer levels in older medical patients presenting to the ED. I commend the authors for conducting a prospective, clinician-blinded study with complete follow-up, and for focusing not only on VTE but also on sepsis, ischemia, and occult malignancy. The findings are clinically relevant, and I believe this work will be of interest to both clinicians and researchers.

There are, however, several issues that I suggest should be clarified or expanded before the paper can be accepted:

Assay details and units
Please specify the reagent and units (µg/L FEU vs DDU) used on the Sysmex CS-2100i. Clarifying the analytic range and variability will facilitate comparison across studies.
Modeling of cancer risk
The cut-off of 2000 µg/L is interesting, but the reported discrimination (AUC = 0.63) is modest. I recommend presenting D-dimer also as a continuous variable (e.g., log-transformed with splines) and considering multivariable models that adjust for age, sex, comorbidity, infection/sepsis, and admission status. Providing confidence intervals for the interval likelihood ratios in Table 4 would be valuable.
Potential detection bias
Many cancers were diagnosed very soon after the index presentation, often during hospitalization. Please conduct sensitivity analyses excluding cancers detected within 7 or 30 days to explore whether D-dimer truly anticipates later diagnoses. Competing risk analysis (with death as a competing event) would also strengthen your findings.
Sepsis and inflammation
As infection and inflammation are common causes of D-dimer elevation, it would be helpful to clarify how sepsis was defined (qSOFA, SIRS, SOFA) and to discuss whether elevated inflammatory markers might confound the association between D-dimer and cancer.
Definition of ischemia
Please provide more detail on how ischemic events (cardiac, cerebral, mesenteric, peripheral) were defined and diagnosed.
Resource implications
Although D-dimer values were blinded, it would be informative to report the actual use of imaging or additional diagnostic tests by D-dimer strata. This would allow the reader to better appreciate the potential impact on resource utilization if D-dimer were to be acted upon.
Presentation
- Please ensure consistent reporting of units and percentages across the manuscript.
- Adding numbers at risk to the Kaplan–Meier plot would improve interpretability.
- Consider moving the data on cancer types and metastatic stage (currently in the supplement) into the main text for emphasis.

Overall, I believe the study is valuable and that, with these clarifications and minor additional analyses, the manuscript will provide a robust and clinically useful message.

Author Response

Comment 1: Assay details and units
Please specify the reagent and units (µg/L FEU vs DDU) used on the Sysmex CS-2100i. Clarifying the analytic range and variability will facilitate comparison across studies.

Response 1: We appreciate the reviewer’s helpful suggestion. D-dimer levels were measured using the INNOVANCE® D-Dimer assay (Siemens Healthineers, Marburg, Germany) on the Sysmex CS-2100i. Results are reported in µg/L fibrinogen equivalent units (FEU). This information is added to the Methods section of the manuscript (page 3, Line 127-128).

Comment 2: Modeling of cancer risk
The cut-off of 2000 µg/L is interesting, but the reported discrimination (AUC = 0.63) is modest. I recommend presenting D-dimer also as a continuous variable (e.g., log-transformed with splines) and considering multivariable models that adjust for age, sex, comorbidity, infection/sepsis, and admission status. Providing confidence intervals for the interval likelihood ratios in Table 4 would be valuable.

Response 2: Following the reviewer’s suggestion, we additionally analysed D-dimer as a continuous variable. We re-evaluated the association between Ln D-dimer and malignancy using both restricted cubic splines and a linear model. The spline analysis confirmed a linear relationship (non-linearity p = 0.90), and the linear fit overlapped the spline curve (See figure attached). Therefore we used Ln D-dimer in subsequent analysis and adjusted for age, sex, comorbidity (Charlson Comorbidity Index), infection, sepsis, and admission status. In this model, higher D-dimer levels remained independently associated with occult malignancy (adjusted HR 1.90, 95% CI 1.24-2.89, p=0.003, supplemental Table S2). Furthermore, we conducted Fine-Gray competing risk analyses (with death as a competing event) and found similar results (supplemental table S2).

We adjusted the results section and discussion section accordingly (Page 4, Line 152-153, Line 155-157 & Line 163-164); Page 9, Line 236-240 & Line 259-260; and Page 10, Line 296-298).

Confidence intervals for the interval likelihood ratios have been added to Table 4.

Comment 3: Potential detection bias
Many cancers were diagnosed very soon after the index presentation, often during hospitalization. Please conduct sensitivity analyses excluding cancers detected within 7 or 30 days to explore whether D-dimer truly anticipates later diagnoses. Competing risk analysis (with death as a competing event) would also strengthen your findings.

Response 3: We thank the reviewer for this insightful comment. We performed additional sensitivity analyses excluding patients with an occult malignancy diagnosis within 30 days of the index presentation. We performed the Cox regression analyses using both log-transformed D-dimer and D-dimer with cut-off. Furthermore, we conducted Fine-Gray competing risk analyses (with death as a competing event). The direction and magnitude of the associations were consistent with the main analyses, although the number of occult malignancy events after exclusion of early cases was limited, resulting in wider confidence intervals for D-dimer >2000 (Supplemental Table S3).

We adjusted the results section and discussion section accordingly (Page 9, Line 240-242 and Page 10, Line 305-307).

Comment 4: Sepsis and inflammation
As infection and inflammation are common causes of D-dimer elevation, it would be helpful to clarify how sepsis was defined (qSOFA, SIRS, SOFA) and to discuss whether elevated inflammatory markers might confound the association between D-dimer and cancer.

Response 4: We defined sepsis according to the SIRS criteria and/or the qSOFA, which is stated in our Methods section. Although infection could potentially confound the association between D-dimer and occult malignancy (since elevated D-dimer levels in infection are not related to cancer), we performed a multivariable analysis in which D-dimer (both as a continuous variable and dichotomized at the 2000 µg/L cut-off) was adjusted for infection and sepsis. In this analysis infection was negatively associated with malignancy, while D-dimer remained independently predictive with similar HRs (Supplemental Table S2).

Comment 5: Definition of ischemia
Please provide more detail on how ischemic events (cardiac, cerebral, mesenteric, peripheral) were defined and diagnosed.

Response 5: We adjusted the methods section to make this more clear (Page 3, Line 101-105). We wrote: “Ischemia was defined as the occurrence of cerebral, peripheral, myocardial, intestinal, or other organ ischemic events during the ED visit. Events were verified using medical records, based on documented clinical diagnoses and, when available, supporting imaging (e.g., computed tomographic (CT) scans) or laboratory results (e.g., troponin, lactate).”

Comment 6: Resource implications
Although D-dimer values were blinded, it would be informative to report the actual use of imaging or additional diagnostic tests by D-dimer strata. This would allow the reader to better appreciate the potential impact on resource utilization if D-dimer were to be acted upon.

Response 6: Unfortunately, detailed data on imaging or additional diagnostic tests (e.g., CT pulmonary angiography, venous ultrasound) were not collected in this study, we focused on the clinical diagnosis. Therefore, we are unable to report diagnostic test utilization by D-dimer strata.

Comment 7: Presentation

Please ensure consistent reporting of units and percentages across the manuscript.

Response 7a: We carefully reviewed the manuscript and ensured consistent reporting of all units and percentages throughout the text, tables, and figures.

Adding numbers at risk to the Kaplan–Meier plot would improve interpretability.

Response 7b: We adjusted the Kaplan-Meier plot with the numbers at risk.

Consider moving the data on cancer types and metastatic stage (currently in the supplement) into the main text for emphasis.

Response 7c: We considered moving the data on cancer types and metastatic stage to the main text; however, we believe that keeping these data in the Supplement provides a clearer and more concise presentation of the main results, while still making the detailed information available to interested readers.

Overall, I believe the study is valuable and that, with these clarifications and minor additional analyses, the manuscript will provide a robust and clinically useful message.

We thank the reviewer for the positive feedback and are pleased that the study is considered valuable and clinically useful.