Imaging Characteristics and Prognostic Value of Isolated Pulmonary Metastasis from Colorectal Cancer Demonstrated with18F-FDG PET/CT

Objective Solitary pulmonary lesions (SPNs) in patients with a history of colorectal cancer (CRC) may be attributed to metastatic lung tumors, primary lung cancer, or benign nodules. We aimed to analyze the imaging characteristics of SPNs in CRC patients to differentiate these pulmonary nodules and evaluate the prognostic value of isolated pulmonary metastasis from CRC using 18F-FDG PET/CT. Methods From January 2013 to January 2021, 62 CRC patients with SPNs demonstrated with 18F-FDG PET/CT were retrospectively enrolled in the present study. We compared the radiological and clinical characteristics of these patients. In addition, survival time and prognostic factors were statistically analyzed using the Kaplan-Meier method and multivariable Cox proportional hazards models. Results There were 33 cases of isolated lung metastasis, 20 cases of second primary lung cancer (SPLC), and nine cases of benign nodules. The proportion of nodules with a maximal diameter greater than the median value was lower in the isolated lung metastasis group compared with the SPLC group (p < 0.05), showing polygonal shape, ill-defined margin, pleural indentation, air bronchogram, speculation, and ground-glass opacity. Patients with isolated lung metastasis had a significantly higher maximal diameter of lung lesion, SUVmax of lung lesion, and 18F-FDG uptake compared with the benign nodule group (p < 0.05). Multivariate analysis revealed that the following two factors were significant independent predictors of PLC: air bronchogram (hazard ratio [HR] =22.327; 95% confidence interval [CI]: 1.910-261.061; p = 0.013) and spiculation (HR =6.148; 95% CI 1.469-25.725; p = 0.013). Initial TNM stage IV (HR =19.831, 95% CI 1.061-370.782; p = 0.046) was extremely associated with a decreased lifespan of CRC patients with isolated lung metastasis. Conclusions The result showed that CT features, including air bronchogram and spiculated margins, could be used to differentiate SPLC from single isolated lung metastasis in CRC patients. In patients with isolated lung metastasis, primary CRC TNM stage IV was associated with a poorer prognosis, and patients with such conditions might need more care.


Introduction
Colorectal cancer (CRC) ranks third in incidence and is the second leading cause of cancer-related mortality worldwide. Over 1.8 million new cases of CRC and 881,000 deaths are estimated to occur in 2018 [1]. Metastasis significantly contributes to treatment failure and cancer-associated death in CRC [2,3]. Pulmonary metastasis is one of the most frequent extra-abdominal sites of CRC [4]. However, about 10% of pulmonary metastasis is presented as solitary pulmo-nary lesions (SPNs), which mimic primary lung cancer (PLC) [5]. Although surgery is always recommended for solitary pulmonary metastasis and PLC, the type of resection procedure may vary from limited resections in cases of solitary metastasis to extended resections with lymphadenectomy if the lesion is PLC [6,7]. 18 F-fluorodeoxyglucose ( 18 F-FDG) positron emission tomography (PET) in combination with CT has become a well-established imaging modality to characterize SPNs and improve CRC staging [8,9]. However, the differences In the present study, we aimed to analyze the imaging characteristics of SPNs in CRC patients to differentiate these pulmonary nodules and evaluate the prognostic value of isolated pulmonary metastasis from CRC demonstrated with 18 F-FDG PET/CT.

Patient Selection.
This study was approved by the institutional review board of the First Affiliated Hospital of Soochow University. Trial registration number: ChiCTR2100045115. Since the trial was a retrospective study, written informed consent for this study was waived by the ethics committee, and no personal information was disclosed. This study complied with the Declaration of Helsinki.
From January 2013 to January 2021, CRC patients with SPNs demonstrated with 18 F-FDG PET/CT were retrospectively enrolled in the present study. The following patients were excluded: (1) patients had additional malignancies; (2) digital image data unavailable for retrospective analysis; (3) patients with appendiceal tumors; and (4) PET was performed in 6 weeks after the end of chemotherapy, and 3 months after any radiotherapy. If the size of the lesion was less than 5 mm, periodical CT surveillance was recommended until lesion growth was observed.
2.3. Image Acquisition. PET/CT images were acquired at 60 ± 5 min after injection of 18 F-FDG (dose 0.12 mCi/kg) according to the European Association of Nuclear Medicine (EANM) guidelines [10]. All patients were subjected to food deprivation for 6 h before the 18F-FDG PET/CT (General Electric Medical Systems, Milwaukee, WI, USA). The baseline blood glucose level was lower than 11 mmol/L. Attenuationcorrected images were acquired via low-dose CT (140 kV; 120 mA). Emission PET data were obtained after CT scanning, with 2-3 min/bed position. PET and CT images were reconstructed using a standard iterative algorithm (ordered-subset expectation maximization). Thin-section CT (TSCT) parameters were set as follows: 140 kV, 120 mA, 0.8 s rotation time thickness, 700 mm field of view, and 3.75 mm slice thickness. An ADW4.1 workstation (GE Healthcare) was used to display whole-body PET/CT images. 18 F-FDG PET/CT images were independently evaluated by two experienced nuclear medicine physicians who were blinded to the clinical information of all the subjects, and the results were recorded by consensus. 18 F-FDG uptake was evaluated using SUV max , and the highest FDG uptake was considered the SUV max of the lung lesion. SUV max ≥ median was classified as PET-positive.

Statistical Analysis.
All statistical analyses were performed using the GraphPad Prism 5.0 software (GraphPad Software Inc., San Diego, CA, USA) and IBM SPSS 19.0 software (SPSS Inc., Chicago, IL, USA). ITP was defined as the period between the diagnosis of CRC and the identification of SPNs. OS was defined as the duration between the diagnosis and the final follow-up or all-cause mortality. The clinicopathological characteristics of the groups were compared using Pearson's chi-square test. The survival rates were estimated with the Kaplan-Meier method, and differences among groups were evaluated by a log-rank test. Univariate analysis for OS-associated factors was performed. Risk factors with statistical significance upon univariate analysis were introduced into a Cox proportional hazards model for multivariate analysis. A p value of <0.05 was considered statistically significant.    Figure 1). The maximal diameters of the nodules, lesion location, SUV max of lung lesion, 18 F-FDG uptake, calcification, cavity, and presence of enlarged mediastinal lymph nodes, or other sites of metastasis were not significantly different between the patients with isolated lung metastasis and those with SPLC (p > 0:05). However, the proportion of nodules with a maximal diameter greater than the median value was lower in the isolated lung metastasis group compared with the SPLC group (p < 0:05), showing polygonal shape, illdefined margin, pleural indentation, air bronchogram, speculation, and ground-glass opacity.

PET/CT Imaging Features of SPNs.
We also compared the radiological characteristics of CRC patients with isolated lung metastasis and benign nodules. Patients with isolated lung metastasis had a significantly higher maximal diameter of lung lesion, SUV max of lung lesion, and 18 F-FDG uptake (p < 0:05). GGO component was more frequently detected in benign nodules compared with SPNs. Table 3 shows the results of multivariate analysis for the clinical and radiological factors that could discriminate between isolated lung metastasis and SPLC. Multivariate analysis revealed that the following two factors were significant independent predictors of PLC: air bronchogram (hazard ratio [HR] =22.327; 95% confidence interval [CI]: 1.910-261.061; p = 0:013) and spiculation (HR =6.148; 95% CI 1.469-25.725; p = 0:013) (Figure 2). Table 4 demonstrates that the maximal diameter and the 18 F-FDG uptake of lung lesions did not affect the prognosis of CRC patients with isolated lung metastasis. Adjuvant chemotherapy of CRC was associated with a better prognosis, while hilar and/or mediastinal lymph node metastasis, air bronchogram component, initial TNM stage IV, CA19-9 level, and ITP<12 months were all significantly related to a worse prognosis in CRC patients with isolated lung metastasis ( Figure S1). Furthermore, the multivariate Cox proportional hazards model indicated that initial TNM stage IV (HR =19.831, 95% CI 1.061-370.782; p = 0:046) significantly contributed to the reduced lifespan of CRC patients with isolated lung metastasis ( Figure S2).

Discussion
Isolated lung metastasis is more frequently reported in CRC patients than in those with other extrathoracic malignancies [5,11,12]. It is sometimes difficult to determine whether an SPN is PLC or an isolated lung metastasis because SPLC occasionally mimics isolated lung metastasis of CRC patients [13,14]. However, different surgical strategies are adopted to treat these two lung nodules. In general, the gold standard for SPLC is the complete surgical resection with lobectomy and mediastinal lymph node dissection. In contrast, the treatment of lung metastasis is minimally invasive surgical resection to preserve healthy lung parenchyma in case a repeat operation is needed [15,16]. Therefore, it is essential to distinguish between PLC and isolated lung metastasis before surgical planning.
In the present study, we investigated the feasibility of using PET/CT to differentiate PLC and isolated lung metastatic lesions in CRC patients. Air bronchogram and spiculated margins were beneficial discriminatory factors in multivariate analyses, which was in line with some previous studies [17][18][19][20]. As we know, a smooth or well-defined margin is more common in metastatic nodules than an irregular margin, while PLC usually presents with a non-smooth margin, especially a speculated margin [21,22]. However, the 5 BioMed Research International margin of a nodule is more irregular when the size of isolated lung metastasis is increased [23]. Ohtaki et al. have reported that the presence or absence of GGO and pleural indentation can be helpful factors to distinguish between PLC and lung metastasis from CRC [24]. In their study, patients with more than one SPN are also included. Similarly, the proportion of GGO and pleural indentation in the isolated lung metastasis group was significantly higher compared with the PLC group in our univariate analysis. However, in multivariate analysis, they were not significant factors. A possible selection bias of patients suggested that these results should be interpreted with caution.
In terms of the prognosis of CRC patients with pulmonary tumors, survival differences were observed in the present work. Our data showed that OS was not significantly different between the isolated lung metastasis group and the PLC group. Chang et al. have reported that the median OS of CRC patients with PLC is longer than that of CRC patients with metastatic lung lesions, while the betweengroup difference is not significant. They believe that CRC patients with SPLC are diagnosed in the early clinical stage because of regular surveillance imaging. Early detection and immediate surgical resection may prolong survival and improve prognosis [25]. However, Ohtaki et al. have found that the outcome of CRC patients with lung metastasis is sig-nificantly better than the patients with radiologically solid lung cancer [24]. In their opinion, the introduction of less invasive surgical techniques, perioperative management, and various types of chemotherapy has improved the survival of CRC patients with lung metastasis. This controversy may be attributed to the selection bias of patients. In the present study, only CRC patients with SPNs were analyzed.
Consistent with previous reports [26,27], we also found that the TNM stage of primary CRC significantly impacted survival. Nonetheless, there was no definite agreement on the value of its prognostic role. Furthermore, some authors fail to confirm a correlation between the primary TNM stage and survival [28,29]. This discrepancy may be explained by different criteria for patient selection and treatment of primary CRC. 18 F-FDG PET/CT has been shown to have high sensitivity and specificity for detecting mediastinal and hilar lymph node metastases [30]. According to our data, patients with concomitant mediastinal and/or hilar lymph node metastases are more likely to be diagnosed with PLC than metastatic lung tumors.
In the present study, we also analyzed the prognostic values of factors about PET/CT imaging, such as metastasis detected on PET/CT, maximal diameter of lung lesion, and SUV max . Only hilar and/or mediastinal lymph node  BioMed Research International metastasis were related to a worse prognosis in CRC patients with isolated lung metastasis in univariate analyses. Other factors were not found to be significant in this study. Moreover, some previous studies have shown that the presence of hilar or mediastinal lymph node metastases can predict a poor prognosis for CRC patients with lung metastasis [27,28], and our results supported these results. This study has several limitations. First, this was a retrospective study with a relatively small sample size. To the best of our knowledge, although this study was the largest singlecenter cohort of CRC patients with SPNs, the number of the cases was still small. Because the incidence of SPLC and single isolated lung metastasis in CRC patients is relatively low, such small sample numbers may lead to difficulty in statistical comparison. A prospective study and a larger cohort are needed to confirm our results. Second, the median follow-up time was 20 months, which was inadequate. More than 5 years of follow-up is required to confirm the prognostic value of isolated pulmonary metastasis from CRC demonstrated with 18 F-FDG PET/CT. Third, CT features were visually analyzed in this study, which might raise the possibility of inter-observer and intra-observer variability despite consensus reading. More quantitative analysis tools, such as radiomics, would be more helpful for a more accurate interpretation. Fourth, the range of tumor size in this study was relatively large. On the one hand, small tumors are usually featured with well-defined margins, which are easily differ-entiated. On the other hand, small tumors may cause the inaccuracy of SUV max value because of the degree of respiratory motion.
In conclusion, CT features, including air bronchogram and spiculated margins, could be used to differentiate between SPLC and single isolated lung metastasis in CRC patients. In patients with isolated lung metastasis, primary CRC TNM stage IV was associated with a poorer prognosis, and patients with such conditions might need more care.

Data Availability
The datasets generated or analyzed during the current study are not publicly available in order to preserve patient confidentiality but are available from the corresponding author on reasonable request.

Conflicts of Interest
The authors declare that they have no conflicts of interest.

Authors' Contributions
Yu Yu, Jing Zhu, and Yeye Zhou contributed equally to this work.