Time from first tumor manifestation to diagnosis in patients with GEP-NET

Abstract Patients with neuroendocrine tumors (NET) often go through a long phase between onset of symptoms and initial diagnosis. Assessment of time to diagnosis and pre-clinical pathway in patients with gastroenteropancreatic NET (GEP-NET) with regard to metastases and symptoms. Retrospective analysis of patients with GEP-NET at a tertiary referral center from 1984 to 2019; inclusion criteria: Patients ≥18 years, diagnosis of GEP-NET; statistical analysis using non-parametrical methods. Four hundred eighty-six patients with 488 tumors were identified; median age at first diagnosis (478/486, 8 unknown) was 59 years; 52.9% male patients. Pancreatic NET: 143/488 tumors; 29.3%; small intestinal NET: 145/488 tumors, 29.7%. 128/303 patients (42.2%) showed NET specific and 122/486 (25%) patients other tumor-specific symptoms. 222/279 patients had distant metastases at initial diagnosis (187/222 liver metastases). 154/488 (31.6%) of GEP-NET were incidental findings. Median time from tumor manifestation (e.g., symptoms related to NET) to initial diagnosis across all entities was 19.5 (95% CI: 12–28) days. No significant difference in patients with or without distant metastases (median 73 vs 105 days, P = .42). A large proportion of GEP-NET are incidental findings and only about half of all patients are symptomatic at the time of diagnosis. We did not find a significant influence of the presence of metastases on time to diagnosis, which shows a large variability with a median of <30 days.


Introduction
Neuroendocrine tumors (NET) are a group of rare and heterogenic tumors that can arise in different organs throughout the body, predominantly in lung, pancreas, and small intestine. [1,2] Overall incidence in Europe is 2.5/100,000 per year, [2] making it an orphan disease. [3] NET occurring in stomach, small intestine, or the pancreas are often summarized as gastroenteropancreatic NET (GEP-NET). [1] In general, different items, predominantly their organ of origin, morphological features, and their grading, classify NET. [4][5][6] Grading (G1-G3) is deducted from a positive immunohistochemistry (IHC) staining for the proliferation marker Ki67. Other important IHC markers in NET are chromogranin and synaptophysin as well as other markers to determine the originating organ, which is important for the choice of treatment. [7] The WHO classification for NET was changed in 2017, which, among others, harmonized different systems used for small intestinal NET (siNET) and pancreatic NET (pNET) as well as changed the threshold for G1 NET from Ki67 <2% to <3%. [5] Most NET are G1 tumors, which are slowly growing and in a large proportion of patients not detected before a metastasized stage. Besides, patients often do not show any deterioration of their general condition and therefore do not seek medical advice in early stages. [8][9][10] If patients show symptoms, these are often unspecific and misleading, which might delay diagnosis in many patients. [4] About 20% of all patients experience carcinoid syndrome, which consists of tachycardia, flushing, and diarrhea, either alone or in combination. [11,12] However, these symptoms are not restricted to NET and do not always trigger further targeted diagnostic measures. Sometimes, patients and treating physicians misinterpret them as, for example, symptoms of menopause in women, irritable bowel syndrome, or to be psychosomatic. [9] As a result, patients are reported to often experience a long time from onset of first symptoms to definite diagnosis, which might influence patients' survival. [1,9]   The aim of our study was to analyze, in a large cohort of patients with GEP-NET from a university clinic, the patients' pretherapeutic pathways from onset of first symptoms to diagnosis, taking into account NET-specific and unspecific symptoms, and diagnostic measures that detected NET.

Study design and ethics
The present retrospective, single center study was performed to investigate the time and pathway to diagnosis for patients with GEP-NET. The study was approved by the institutional review board (internal reference numbers 319/16 and SGI-1-2019) of the University Hospital Frankfurt. Informed consent to participate in the tumor documentation registry was obtained from all patients alive. The informed consent explicitly includes a passage about scientific use. Inclusion criteria of the study were diagnosis with GEP-NET and age ≥18 years. NET with unknown primary tumor were also included.

Patient data
The study database was based on the local electronic hospital charts and was transferred to the local tumor documentation system (Giessener Tumordokumentationssystem). A specific NET dataset was designed and used for documentation of all patients. The dataset included epidemiological and clinical data and is explained in detail in Table S1, Supplemental Digital Content, http://links.lww.com/MD/G408. NET-specific symptoms were defined as diarrhea, abdominal pain, flushing, tachycardia, and carcinoid syndrome (if mentioned specifically without further information), although certainly an overlap between NET specific and non-specific symptoms has to be assumed. Data closure and end of follow-up was February 11, 2019. Table S2, Supplemental Digital Content, http://links.lww. com/MD/G409 contains the anonymized date set.

Statistical analyses
Statistical analyses were performed according to international standards and have been described by us and others before. [12] Categorical variables were described in frequencies and percentages. Continuous variables were represented as a median, IQR, and its range. Continuous variables were compared using the Wilcoxon-Mann-Whitney U test. Contingency table was analyzed by chi-square or Fisher exact test, as appropriate. All tests were 2-sided and P values .05 were considered statistically significant. Analysis was done using International Business Machines Corporation (IBM) Statistical Package for the Social Sciences (SPSS) for Windows (version 22.0; IBM, Chicago, IL), BiAS (version 11, Frankfurt, Germany), and R (R Core Team (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria).

Demographics
486/1030 NET patients in the database fulfilled the inclusion criteria and were included in the study. Further details on patients' characteristics are lined out in Figure 1

Discussion
NET are a rare disease and diagnosis is anecdotally reported to be delayed sometimes by years. We analyzed a large single center dataset to further clarify symptoms before diagnosis of the tumor and the period until the NET is diagnosed.
Across all entities, median time to diagnosis in our left-skewed data was only 19 days with a wide range. A delay in diagnosis of more than half a year occurred in only about one fifth of our evaluable patients, and 95% of all patients are diagnosed within 5 years. Hence, our data indicate a shorter interval between first tumor manifestation and diagnosis as reported in the literature, regardless of the presence of metastases at initial diagnosis. Incidental findings occurred more often in patients without metastases; in other words, metastatic patients were more frequently diagnosed because of their symptoms.
Assessing time to diagnosis in patients, in general, is difficult: Some tumors are diagnosed as incidental findings, for example, during surgery, endoscopy, or imaging for other causes, which leads to a time to diagnosis of 0 up to a few days. Other patients retrospectively report symptoms when their medical history is taken at initial diagnosis, and these might be NET specific, such  as flushing or diarrhea, or unspecific, such as weight loss. However, even NET "specific" symptoms might arise from other conditions and diseases, or patients might have a varying awareness for their symptoms. Hence, determining the real time from onset of first symptoms to definite diagnosis seems difficult and biased.
One large global survey on patient-reported outcomes was undertaken in 2014 and found a mean time to diagnosis of 59 months for US GEP-NET patients [13] and 52 months for the global population [10] across all entities. For 29% of all patients of the global population, time to diagnosis was longer than 5 years. Another survey found a mean time from symptom onset to diagnosis of 36 months for siNET and 24 months for pNETs. [14,15] Both datasets are based on patient-reported outcomes and did not include a review of the patients' files. Also, there is no detailed information as to whether the time distribution was skewed towards shorter or longer periods between symptom onset and definite diagnosis. If longer periods to diagnosis are outliers, median instead of mean time to diagnosis might be a more realistic parameter.
We took a different approach by reviewing the patients' files and collecting all information that was available at primary diagnosis. Thereby, we were able to better exclude bias due to a false or incomplete recall by the patient. However, we had complete information only for about 60% of our patients. Also, we divided the cohort into different sub-cohorts by type of primary diagnosis, primary tumor site, and whether the patient was symptomatic or not.
Demographics and distribution of histology in our cohort were as expected from the literature [16][17][18] ; therefore, we assume that we analyzed a representative cohort and that the results could be transferrable to other tertiary centers.
Considering the difficulties in diagnosing the tumor early, we also tried to better understand the pathway to diagnosis for our patients. We were not able to get information on the number of contacts with a physician or health care provider before diagnosis as reported by Singh et al [10] : however, we found that patients see different referring instances and clinics before a definite diagnosis, which underscores the importance of a family doctor or GP as a coordinating instance. Also, since a majority of patients shows liver metastasis at initial diagnosis, routine ultrasound examinations might facilitate early diagnosis.
Limitations of our study are the retrospective design and the large time span with partly incomplete datasets.

Conclusion
Taken together, we were not able to confirm the reported massive delay in diagnosis for the majority of patients. Although it would be desirable to detect tumors in a not metastasized stage in order Table 3 Metastatic patients with or without incidental findings.  to be able to perform surgery, there are only rarely "red flag" symptoms such as ileus or massive carcinoid syndrome that should have prompted further diagnostic measures. Health care providers should consider NET as a differential diagnosis in patients with unknown findings in endoscopy or imaging. In patients with carcinoid syndrome, which occurs only in metastasized patients, an abdominal ultrasound could detect liver metastases early.